Any attorneys here that can discuss the nuances involved in knowingly lying to the court, which I believe carries penalties for the attorney? And why / whether that would or wouldn’t apply here?
I'm not a lawyer, but I'm pretty sure that's not a thing. Lying to the court is a routine part of litigation, and there is no penalty or sanction for it - especially not against the lawyers themselves.
> I'm not a lawyer, but I'm pretty sure that's not a thing. Lying to the court is a routine part of litigation,
No, its not.
> and there is no penalty or sanction for it
Yes, there is.
> - especially not against the lawyers themselves.
There is a penalty especially against the lawyers. Heck, lawyers have been suspended from practice for lying to courts outside of their capacity as lawyers.
I don't agree that outright lying is tolerated, but one thing I was very disappointed to learn after independently learning about most of the informal logical fallacies is that lawyers do not avoid them. In fact, they use them to win cases.
This is tantamount to, but NOT equivalent to, "lying" in my book. Some of the most famous defenses in history, such as Cochran's "if it doesn't fit, you must acquit" https://www.youtube.com/watch?v=NH-VuP_5cA4 were fundamentally, fallacious appeals in nature.
Hugely disappointing to me. Would love for a lawyer to chime in on this.
If I could come up with any defense of this practice at all, I'd say that if you took my reasoning to its logical extreme, then the persuasive personality of an attorney themselves could be considered a fallacy, and that would be an unreasonable expectation to satisfy the elimination of. You could also argue that if 2 opposing lawyers are both permitted to make fallacious appeals that wouldn't be called out by judges or jury, then in theory they'd mostly cancel each other out. Maybe.
Lying in court is not a crime. You need to be under oath to be able to commit perjury, e.g. by being a witness who is giving testimony which in turn will be used as evidence. Everyone else can lie as much as they want without fear of repercussions.
Did noone here read the article? It says the defendants were the ones who were lying. The attorneys just tried to cast doubt on the evidence, which was at best exacerbated by the defendant's claims.
It definitely carries penalties for the attorney. Each US state has passed 'rules of professional conduct' that govern rules that apply to barred attorneys. Lying to a tribunal, or lying to your client, is illegal and you can be disbarred and fined for it.
Haven't read this case closely but it sounds to me like the attorneys said "it's possible these are deep fakes" and not "these are deepfakes." The former seems like a reasonable argument to make, but according to the judge, a flawed one.
From my understanding, (good) lawyers will throw every possible defense at the wall (within reason) because one might stick, and if you don't do it at the beginning, you can't call a whoopsiedoodles and go back and try again. This is why they always begin with a petition that the venue is wrong and the court doesn't have jurisdiction, if they want to argue that.
That's generally true: you "waive" certain defenses if you don't state them, at certain points in the litigation process.
That said, at the most important moments, typically lawyers will focus on the 3-or-so 'best' (in their view) defenses and focus on those. Then if you lose, you can appeal and say the judge didn't consider your 15th-best defense closely enough :)
> Haven't read this case closely but it sounds to me like the attorneys said "it's possible these are deep fakes" and not "these are deepfakes." The former seems like a reasonable argument to make, but according to the judge, a flawed one.
IANAL, but this sounds appropriate to me? Couldn't all forms evidence be hit with various types of "this might be fake" claims? So I think it's fair that the court assert that you must claim "this is fake", then be willing to battle it out with the opposing side. Otherwise it's not a defense, it's just empty words, isn't it?
Dishonesty toward the tribunal is a violation of the rules of professional responsibility/conduct (model rules 3.3 and 8.4(d), the latter is what they disbarred Clinton for).
Federal courts have rule 11, where an attorney can essentially call out the opposing counsel after some cure period and ask the court to impose sanctions on the attorney that's getting out of line. State courts have some equivalent.
In practice, most lawyers don't lie to the court because their career is dependent on maintaining a working relationship with the courts in which they appear. Even within the same case, if the presiding judge has good reason to believe you're a liar, every subsequent motion or argument you make is going to be viewed with skepticism.
> "Their position is that because Mr. Musk is famous and might be more of a target for deep fakes, his public statements are immune," she wrote. "In other words, Mr. Musk, and others in his position, can simply say whatever they like in the public domain, then hide behind the potential for their recorded statements being a deep fake to avoid taking ownership of what they did actually say and do. The Court is unwilling to set such a precedent by condoning Tesla's approach here."
This seems like the main point - it's not that court won't accept a deepfake defense, but you would have to prove that it was faked, just like any other convincing evidence
It also depends on the court in question. Ie in my recent criminal Jury experience; if i were to pretend deepfakes were used then it wouldn't be on the prosecution to prove that the video is without a doubt real, but rather that a "reasonable person" would believe that the video is real. Which is quite open to interpretation.
For a Jury i think it would come down to how convincingly the defense argued that the video is a deepfake. Just saying the words "it's a deepfake" hypothetically wouldn't/shouldn't be enough to convince a "reasonable person" to believe all video evidence is now fake without concrete evidence (if any is even possible) from the prosecution.
Disclaimer: I'm quite unfamiliar with the nuance of the legal system. But i keep using the words "reasonable person" because this was hammered into us in the Jury. We were given a small book of legal definitions related to the criminal case and just about every page framed each definition around a "reasonable person". This seemed to be how the state defined any nuance between the English language and strict logic.
This would, i imagine, help mitigate a hypothetical defense of "all evidence is faked" or w/e.
I think the reasonable person standard is used for evaluating whether someone's behavior is okay. Like, if you tased someone because they were suspicious, the question would be whether a reasonable person would conclude they were suspicious based on what they did.
Yeah, reasonable doubt applies in "well maybe aliens completely replaced the body and mind of this person before he did this thing recorded on fifteen different phones."
Arguing that a video randomly was deepfaked would go a lot further if you could show other evidence that such was happening.
Nope, i'm thinking about "reasonable person". I was just in the case (a few weeks ago) so this is quite fresh in my mind.
For additional context, the criminal case i was part of the Jury for was handed a book of definitions for the given case. Things like Assault had qualifiers of Self Defense and Stand Your Ground. Each of these little nuance-y definitions included very implicit definitions of say, the force you use in response to assault for self defense or stand your ground. This responsive force was instructed to be judged against something like "Force that a reasonable person would not believe to be in excess based on the attack".
Ie "reasonable" was only able to be decided by a hypothetical "reasonable person". Whether something is reasonable or not was only applicable in the context of the person who is deciding if it's reasonable, also themselves being a reasonable person.
Just about every definition page had an explanation quite similar to this. I think due to the fact that human definitions are very difficult to define. Did you hit them too hard? Well we don't have pounds per square inch of your delivery, or w/e concrete definition could use, so the legal definitions were bounded by this hypothetical "reasonable person".
It was interesting.
edit: I should add, in case it's not clear, that we the Jury were instructed by the judge on the legal definitions. That was the book. We were to view the case through the lens of these legal definitions with the evidence presented. What specifically we were deciding was very important and meticulously managed by the Judge.
Yea, and i'm not making any grand statements or anything. Just what my experience was. I as a Juror felt woefully under prepared. It was kinda alarming, to be honest. They explicitly didn't want you to research the law even during the process, so they did want you (me) under prepared and "fresh".. but that didn't really mitigate the unease.
The jury is supposed to make all the decisions with respect to weighing evidence once it’s admitted. The standard for admitting evidence is very low if it can be authenticated, and it’s basically “is it relevant and does it have an impact on the questions the jury will be asked to ultimately come to a conclusion on without unduly prejudicing the jury in some way.” The jury’s decision must meet the standard of liability (more likely than not, clear and convincing, no reasonable doubt, etc.)
So it’s entirely possible for a ton of evidence to come in that none of it meets what you’d consider the “no reasonable doubt” standard and for a jury to consider all that suspicious evidence to be true enough that they can convict someone under the “no reasonable doubt” standard.
> Doesn't this also mean that a deepfake presented as evidence of crime makes you guilty until proven innocent?
It means the same thing that's always been true remains true. If a prosecutor (or one side in a civil case) presents fake evidence that is convincing to the court, the other side will have to prove the evidence to have been faked. Adding machine learning to the equation and calling the fakes "deep" doesn't change that basic reality.
Consider everyone who has ever had drugs/weapons planted on them by the cops. As a society we assume most of the people claiming it happened are lying, but some of them are telling the truth and some of that subset actually manage to prove it (usually because the cops managed to record themselves doing it). If the authority figures trusted by the courts intentionally present false evidence then you are stuck not having to prove your innocence but having to prove the evidence to be false or at least untrustworthy.
But then the defense has to pay the cost of hiring an expert to prove it's not authentic.
It's really hard to predict where the technology is headed with regards to both generating and proving the non-authenticity of deepfakes, but it is concerning that this could open up people without money to pay for defense to be ruined by people with money to pay for deepfakes.
Yes. Which he's not gonna do unless he wants a perjury charge, because it's a seven year old video on a reputable channel with multiple well-known reporters interviewing him.
Yes but people without access to good enough lawyers are going to get the short end of the stick as always, and get framed for something they didn't do. The number of innocent people in jail could skyrocket.
The new thing is that fakes that could convince an ordinary person that they're real were in the realm of state intelligence agencies, and now they're in the realm of anyone with a computer. Which causes the probability that a given video is fake to go from negligible to significant.
It's also not clear if experts always can tell if something is a fake or not, if it was created by an algorithm which is less than a year old or doesn't happen to have a lot of the artifacts that these algorithms sometimes introduce.
On topic though, this doesn't make "It could be fake," a reasonable defense. Sure, the surface area will get larger, but we have that problem across all digital media at this moment. Deepfakes are no longer unique.
The courts will have to deal with it one way or another, and right now expert witnesses is the tool they have, and the tool defendants will need to use.
> On topic though, this doesn't make "It could be fake," a reasonable defense.
Why not?
If you go into court claiming you have a contract with someone but you have nothing notarized or even signed, you're the one who has to provide evidence of your claim. A video is no longer sufficient to do that because any plaintiff or internet troll now has the ability to convincingly fabricate one. It's no more proof of anything than the plaintiff making an unsubstantiated claim.
> The courts will have to deal with it one way or another, and right now expert witnesses is the tool they have, and the tool defendants will need to use.
Why should it be the defendants who have to pay for the expert instead of the plaintiffs?
The difference is the difference between "it may be fake" and "it is fake". "May" being the key word that makes it a meaningless observation instead of a defense. If someone wants to defend themselves against video evidence, they need to say that it's fake, and create reasonable doubt about its veracity. You don't do that with "maybe" or "could be".
> Why should it be the defendants who have to pay for the expert instead of the plaintiffs?
TL;DR: that's how our legal system works in the US, for better or worse.
EDIT: You probably don't have to attack the video as a deepfake directly, either. "That's not me, I wasn't at X, I was at Y, and here are receipts and witnesses" are highly likely to work too. Basically, an alibi.
Suppose the video isn't of an old statement made by the defendant which they don't remember making, it's a video of one of their products catching fire sourced from an anonymous YouTube account. They have no way to know if that video is fake or not, so on what basis could they claim to know for sure that it is?
But because it could very plausibly be fake, they should be able to demand proof that it isn't from the party trying to introduce it as evidence. For example, by having it authenticated by the person who created it, whose credibility could then be questioned etc.
> "That's not me, I wasn't at X, I was at Y, and here are receipts and witnesses" are highly likely to work too. Basically, an alibi.
An alibi is the defense proving that they couldn't have done it. The burden of proof is supposed to be on the plaintiff. Creating deepfakes is so easy that a plaintiff in possession of one by itself proves nothing.
Isn't this backwards in criminal court? Wouldn't the onus be on the prosecution to prove that the video was real?
This sounds ridiculous since we lived most of our lives in the video-is-real era that has sort of been coming to an end. I recall the Kyle Rittenhouse trial had an interesting take on this where the defense asked to throw out upscaled video which contrary to every police procedural TV show is a lot like just fabricating data. With the amount of post processing and AI a cell phone camera does these days you could argue everything they pump out is doctored and fake by default because it kind of is. And most video is stored with lossy algorithms even without the post processing.
The onus on the prosecution/plaintiff is to prove the defense guilty at whatever standard of evidence is required for the case. Not to prove or disprove any claims about evidence. If the defense makes a claim about any piece of evidence, it is up to them to prove to the court's satisfaction that their claim is truthful.
The burden is on whichever side is attempting to introduce a piece of evidence.
From the federal rules of evidence:
> (a) In General. To satisfy the requirement of authenticating or identifying an item of evidence, the proponent must produce evidence sufficient to support a finding that the item is what the proponent claims it is.
Unfortunately, the rules of evidence are pretty vague about what is actually required to prove that a video is admitable as evidence; but the burden to do so is absolutely on the party introducing it.
Those are informative links. Sure. But that doesn't mean they have to prove it wasn't manufactured by martians 50 million years ago just because the other side brought up that possibility just to bring the piece of evidence in to the trial.
The standard for admitting evidence is far below reasonable doubt, even in a criminal case. Generally speaking, all you need is someone to get on the stand and say (under oath), "yes, that video is an accurate copy of the video I took".
If you have that (absent other possible evidentuary concerns), it is up to the trier of fact (e.g. jury) to determine if how much they believe the evidence.
The problem here is the prosecution cannot meet even that low bar, and instead is relying on the various exceptions to the requirement.
Rule 902 lists evidence that is "self authenticating", and videos are not on it.
I think you're reading a little too much into that. I think it's worth comparing it to other forms of evidence. It's incredibly common in civil disputes for one party to give untruthful evidence, sometimes out of malice, sometimes simply due to the human psyche (plenty of evidence of people believing falsehoods after all). Likewise, tampering with physical evidence is far from new - tampered photographs have been going around for over a hundred years, while altered financial records and planted drugs are seen in a court somewhere every day. Practically then, the status quo appears to be that if a prosecutor or litigant has done some due diligence and have a good faith evidence based conclusion that the evidence has not been tampered with (by e.g. having the cameraperson swear under oath that the video is an accurate representation of what they saw), then they will be allowed to present it, and that won't change with the proliferation of deepfake video and audio.
At that point the reliability of the evidence may of course be called into question, as with any evidence or witness. The question mark would be whether it was convincing.
In criminal court at least, evidence is only admitted based on a chain of custody. You're not supposed to enter random videos off YouTube into evidence - you can only enter a piece of video into evidence based on witness testimony about how the video was taken.
As such, the question of deep-fakes should be relatively irrelevant to (criminal) courts. If a witness claims that a particular video was filmed on a particular day in a particular way, that is as much evidence as you need. Whether that witness is lying remains to be proven during the course of the trial.
Things may be different in civil court, where the standards for evidence are more lax I believe.
> In criminal court at least, evidence is only admitted based on a chain of custody.
Prosecution has a freer reign during deposition and can play a video from Youtube[1] and ask the subject of the video if they were at the location at a specific date and time, and if the words are their own.
This is going to be interesting in the world of supporting evidence for criminal cases.
For example, lets say someone accused of a violent crime and the prosecutor wants it upgraded to a hate crime. Now lets the accused did not post a life of hate on facebook. But instead there are just a few audio recordings of the accused saying "I hate X people, I want to kill them all" brought to the court by someone else and not found in the accused possession.
As time progresses it will likely become more difficult to prove that the audio is real. Deepfakes of audio are convincingly good these days. Soon the recordings of politicians saying terrible things will be met with a default reply of "it's a deepfake", and it's likely a bunch of them will be.
The example you described is handled by our current system. (For the moment, let's put aside whether or not the system always works as intended). Evidence isn't just "brought to court by someone else". It doesn't just appear. Someone needs to be saying "this is the accused saying this"- during the discovery process before the trial the defense will have an opportunity to refute that claim. If it's another witness, they will be cross-examined by the defense and the jury will need to decide if the testimony is credible or not. If they lie, they will have committed perjury, and all of what I just said is true (roughly, I'm not a lawyer) whether you use AI or a good voice actor.
It sounds like the defense is trying to disqualify any video of Musk being used at trial on the basis it may be fake.
If any evidence was automatically disqualified on the basis that it merely could be fake, there could never be any successful prosecution.
The defense is still welcome, of course, to make an argument and present evidence that any video or other evidence provided by the prosecution is fake.
This is already done, at least with many modern courts.
Substantiating your evidence has to happen via a pretty specific process, and there is an entire process for entering stuff into evidence with rules regarding chain of custody, validation, etc.
When it comes to evidence, it's not a specific claim against anyone, it's an item that is being introduced to support a claim. That should be validated and checked before it even gets to the court room. If either counsel believes it to be fraudulent, they need to review it and make the claim.
From the article, this was not really the case; instead, a video of a conference with many attendees and participants showed Musk making the claim in question.[0] Tesla's (or Musk's? Not sure) lawyers made a very weak statement:
"[Musk], like many public figures, is the subject of many ‘deepfake’ videos and audio recordings that purport to show him saying and doing things he never actually said or did,” Tesla said"
This is not really an argument towards anything about the video; it doesn't outright say it's a deepfake (likely because it's not), it doesn't provide any evidence to a claim that the video in question is a deepfake, it does nothing to establish who would have produced this deepfake, it doesn't question the YouTube channel owners, Recode, on if they are ready to stand by their statement, etc.
In this instance, the lawyers defending Tesla/Musk are trying to muddy the waters; they aren't bold enough to say Recode made a deepfake, they aren't trying hard to position that the video _is_ a deepfake, they're just trying to sow doubt into the minds of the court about whether to consider normal evidence like recordings or screenshots as valid given that Musk is a public figure. (not sure if there's a jury, but if there is, they want to confuse the jury also)
The judge was right in my opinion to slap the defense counsel on this, as it's transparent and pithy. It was a bad-faith statement to make for a legal team in my opinion as they weren't trying to _say it was a fake_, they just wanted to sow doubt without any proof.
It's even a bit hopeful in my opinion as to me it says the court is holding Musk and the businesses he associates with accountable for Musk's actions and statements; Musk has tried this before when he "joked" about buying Twitter and the courts didn't take on any of the arguments they put up. As I see it, it's another instance of Musk and counsel representing his interests trying desperately to escape culpability.
Edit: Wrong link on first version. I was trying to see if any other channels had coverage of the Code Conference 2016, but looks like it's just Recode.
In this case it's relatively easy to prove the legitimacy of the video. It was published by a real news organization 6 years ago, is a video of an interview conducted by two well known and respected reporters (Kara Swisher and Walt Mossberg), was filmed in front of a live audience (additional witnesses), and it's being hosted on a 3rd party's website for 6 years. There should be piles of evidence that the video is real.
There should be higher degree of burden on the defense than claiming "fake news" and waving your hand.
The burden overall is with the prosecution. Any given piece of evidence can be weighed independently by the jury. For instance, the prosecutor can call 100 eye witnesses and isn't required to prove each is correct beyond the shadow of a doubt. The defense can argue it's a conspiracy, but they will need to supply some evidence, not just assert it and say "you couldn't prove 100 people didn't conspire"
Plenty of evidence accepted in court can be faked. A written ransom note might be faked. A receipt for a gun could be faked.
How believable the emails are will vary. Do you still have the headers? Is there a valid DKIM signature in those headers? Can you bring the people CCed on the email as witnesses? etc.
I don't know what authentication information is stored in an email header that would be difficult to fake. But if the email is stored on a third party's server, the chain of custody and time of creation can be independently documented.
For an old email, it would be up to the opposing counsel to argue that the person who made it made it for the purpose of fabricating evidence however many years ago.
Some image enhancement is scientifically valid. Stacking frames could make a blurry license plate legible. I don't want to live in a world where that is inadmissible.
> Isn't this backwards in criminal court? Wouldn't the onus be on the prosecution to prove that the video was real?
The judge is very much for the prosecution deposing Musk to ask him - under oath - if he can corroborate or dispute what's in the video. The defense was asking the videos to be thrown out entirely because they could be fake.
It's not backwards. The prosecution makes a claim and supports it with evidence. The defense makes the claim that the prosecution's evidence is fake, thus the defense needs to substantiate that claim. That's been true even before deepfakes were a thing.
Interestingly, actually, the "way of weighting the trustworthiness of a piece of evidence" is generally not the job of the court, but of the jury. Assuming the evidence is admissible (meets some minimum standards), the court system is premised on a jury of peers looking at something and deciding if they think it's true or not (such as the testimony of a witness).
Normally in court you have to provide foundation that the evidence is accurate, for video that normally means the person that recorded the video tells the jury it is a true and accurate representation of what they saw.
however the courts have been lack of late allowing police to download social media and YT videos and attest that "the video is a true and accurate representation of what I saw on social media post" because they often can not find, or can not secure the person that actually took the video (or the video was taken by the defendant who can not be made to testify against themselves)
It is very dangerous when we start lessing the requirements for foundation in the law, and the courts have been doing that on many fronts when it comes to "forensic science" (see bite mark evidence as another example of this)
What Musk is doing is sneakier and more deceptive in a different way than that, at an earlier stage. He's trying to get out of having to make any definitive statement about if it's a real video or not.
> The attorneys for Huang’s family sought to depose Musk regarding recorded statements from 2016 in which he allegedly said: “A Model S and Model X, at this point, can drive autonomously with greater safety than a person. Right now.”
> Tesla, however, opposed the request in court filings, arguing that Musk, the Tesla CEO, cannot recall details about the statement and questioning the authenticity of the recording.
The judge's ruling here is "ok then if you want me to think it's fake, tell me that under oath."
It's doubtful Musk would be willing to claim this one was fake - there are a lot of witnesses that could be dug up. He could do the weaker "I don't remember" song and dance, but he's trying to just avoid the issue entirely.
Funny, I got the impression he was basically saying "I'm free to say anything at all now and can no longer be held in any way accountable for my own words."
> He's trying to get out of having to make any definitive statement about if it's a real video or not.
How do you propose to prove if he actually remembers or not?
> It's doubtful Musk would be willing to claim this one was fake - there are a lot of witnesses that could be dug up.
But then the prosecution doesn't need the video because they would have all of these witness, so why should it be admissible if it can't be authenticated?
> How do you propose to prove if he actually remembers or not?
He doesn't have to remember it for it to be good evidence. If he wants to say "I don't think I would've said that", go for it.
> But then the prosecution doesn't need the video because they would have all of these witness, so why should it be admissible if it can't be authenticated?
The video is perfectly good evidence (and its age and provenance make it pretty reputable). If the defense contests its veracity, that's when you start going to witnesses.
> He doesn't have to remember it for it to be good evidence.
He doesn't have to remember it for it to be bad evidence either.
> If he wants to say "I don't think I would've said that", go for it.
But then what are you trying to hold him to? A subjective opinion?
> The video is perfectly good evidence (and its age and provenance make it pretty reputable). If the defense contests its veracity, that's when you start going to witnesses.
The defense will always contest its veracity if the burden is on the prosecution to prove it and they haven't. The point is that they should have to prove it.
> But then what are you trying to hold him to? A subjective opinion?
He needs to make a claim about the evidence in question if he thinks it's fake. (He doesn't. He knows it's a genuine video.) Again, a vague assertion that "sometimes people make deepfakes about me" doesn't cut it.
> The defense will always contest its veracity if the burden is on the prosecution to prove it and they haven't.
The defense is highly unlikely to make the claim that a seven year old video on a prominent channel with top industry reporters didn't happen. Plenty of pieces of evidence get entered into the court's records without being contested.
> He needs to make a claim about the evidence in question if he thinks it's fake.
Why? What if he thinks it could be fake, and doesn't know? Why should that reduce the other party's burden?
> The defense is highly unlikely to make the claim that a seven year old video on a prominent channel with top industry reporters didn't happen.
But it's not the defense claiming that it didn't happen, it's the plaintiff claiming that it did. They should have to prove their claims.
And sometimes they'll be able to.
> Plenty of pieces of evidence get entered into the court's records without being contested.
They don't get contested when the defense doesn't want to aggravate the judge by demanding proof of something they know the other party has, even if they have a right to. That's a different question than whether the other party should have the burden of proof in the cases where they might not be able to meet it.
> What if he thinks it could be fake, and doesn't know?
He will be asked to justify that belief, and that justification will be weighed against the provenance of the video. That he's not saying that is telling; claiming the video is faked would likely hurt his credibility given the number of times he's provably made statements of this kind.
> They should have to prove their claims.
They have submitted it as evidence. The jury weighs if it's credible. If Musk chooses not to rebut/contest it, that's on him, not the plaintiff.
Why? His lawyers could provide the same justification: Convincing deepfakes are easily produced at low cost by adversarial plaintiffs and internet trolls.
> That he's not saying that is telling
People are imperfect and if you misspeak under oath there are consequences. Testifying at all subjects you to cross-examination, which is time consuming and subjects you to lawyer tricks by the other party. Not wanting to do it is the default.
> They have submitted it as evidence. The jury weighs if it's credible.
First the judge decides if it's even admissible. There are some forms of evidence so weak that they only serve to prejudice the jury, and unauthenticated video is pretty much there now.
Demanding that the introducing party authenticate it should be table stakes at this point.
> Why? His lawyers could provide the same justification: Convincing deepfakes are easily produced at low cost by adversarial plaintiffs and internet trolls.
Cool. Video evidence, written evidence, witness testimony, and essentially everything else is now off the table. Congratulations; you've gotten rid of the entire trial court system.
> Testifying at all subjects you to cross-examination, which is time consuming and subjects you to lawyer tricks by the other party. Not wanting to do it is the default.
Yes. If you aren't willing to argue against a piece of evidence in a convincing fashion, you have to live with that decision. The jury gets to draw inference from "huh, they didn't say anything in response to that..."
> Demanding that the introducing party authenticate it should be table stakes at this point.
They'll introduce it with someone describing it something like this:
"This is a seven year old video hosted publicly on YouTube, on the channel of a long-standing reputable organization, of a live event hosted by well-known reporters, which received contemporary news coverage. The claims in the video made by Musk are of a similar nature to others he has demonstrably made."
That's sufficient. Musk's lawyers are free to object, if they've got a good argument against all of that. They're also welcome to try and convince the jury that Musk never made the claims. I suspect the plaintiffs will be able to find plenty of corroborating evidence should it come to that.
> Cool. Video evidence, written evidence, witness testimony, and essentially everything else is now off the table. Congratulations; you've gotten rid of the entire trial court system.
You're going to create a deepfake of witness testimony in front of a live jury? How does that work?
And it's not that you can't submit a video, it's that if you want to, you should have to go find the person who took the video and have them testify that what it shows is what actually happened. And have them be identified, so their credibility can be called into question, e.g. if it turns out they're a long-time antagonist of the defendant.
> Yes. If you aren't willing to argue against a piece of evidence in a convincing fashion, you have to live with that decision. The jury gets to draw inference from "huh, they didn't say anything in response to that..."
But they did have something to say in response to that: Deepfakes are cheap and easy to create. That's a general argument, not a claim about the facts of the case, so it doesn't require testimony to substantiate.
> "This is a seven year old video hosted publicly on YouTube, on the channel of a long-standing reputable organization, of a live event hosted by well-known reporters, which received contemporary news coverage. The claims in the video made by Musk are of a similar nature to others he has demonstrably made."
None of which proves that the video is real. At best it's evidence that it's a possible fake which is seven years old and was published by an established organization.
What they could do is find the reporter who did the interview and bring them in to testify that what the video depicts is what actually happened, if they have a better memory of it than Musk. Which they may very well be able to do. But the point is that they should have to.
> You're going to create a deepfake of witness testimony in front of a live jury? How does that work?
Witness testimony is trivially fabricated. It happens all the time in court. The solution is to rebut it, not ban it.
> But they did have something to say in response to that: Deepfakes are cheap and easy to create.
That's not a reasonable response, any more than "I saw him shoot the person" can't be rebutted with a vague, hand-wavy "well sometimes people lie" by itself.
> None of which proves that the video is real. At best it's evidence that it's a possible fake which is seven years old and was published by an established organization.
> Witness testimony is trivially fabricated. It happens all the time in court. The solution is to rebut it, not ban it.
Witness testimony is under oath. You can lie, but then you can go to jail. And you can be cross-examined, unlike a video. The risk of punishment reduces the incentive to lie and the interactivity increases the potential for it to be discovered.
Witnesses also have varying credibility. If the witness is the plaintiff, they have an obvious incentive to lie and juries know that. Whereas a video from an unauthenticated source is no more credible but fosters the intuition that you're seeing something with your own eyes.
> I don't think you're arguing in good faith.
I think I am.
I think we can agree that a random video from a random YouTube channel has no credibility. But if you're going to stake the value of your evidence on the credibility of the source, you should at least have to establish that it can actually be authenticated by one of these credible parties and not just e.g. some syndicated clip that got published on their channel.
Which, once again, you may be able to do. But you should still have the burden of actually doing it.
they do have the burden of doing that, to the satisfaction of the jury
if all elmu has to say in response is "well maybe it's a deepfake, but I won't say if I think it is", obviously any jury would be hard pressed to suppress their pressing giggles before rejecting such a weak excuse
> But then the prosecution doesn't need the video because they would have all of these witness, so why should it be admissible if it can't be authenticated?
There's not a "prosecution." Did you read the article I linked? This is a civil lawsuit (which also means a different level of evidence required, ultimately) still in fairly early stages. The trial hasn't even started. "Judge Evette Pennypacker tentatively ordered a limited, three-hour deposition where Musk could be asked whether he actually made the statements on the recordings" - there's no ruling of fact yet about if the video is real, the Judge is just saying "this looks like a bullshit excuse to avoid having to talk about your past comments."
The plaintiff is trying to get Musk to give a deposition regarding his claims about self-driving. Musk is trying to not even have to give that testimony. This is exactly that "we need to figure out if this evidence is real or not" process that you say is needed! Musk is just trying to throw that out and say "nah we don't even need it, it could be fake, ya know." The judge seems to think that a 6+ year old video from a fairly well-known publication carries enough weight compared to a lazy "IDK, sometimes people deep fake me" statement to be worth asking Musk further questions. That seems reasonable to me.
> How do you propose to prove if he actually remembers or not?
Obviously that's impossible, but it's also irrelevant. If he says he doesn't remember, but you have a video with an uploaded-to-Youtube-timestamp from 2016, and contemporary coverage, then you say "well, since you don't remember for sure, we're gonna go with all the evidence that suggests you did say this."
"Is it likely, ladies and gentlemen of the jury, that Musk, who has stated multiple times in hundreds of interviews, and company statements, and videos that the company created on websites entirely under their control, would or would not have made a statement materially and structurally similar to this one?"
Also, how good was deep faking in 2016? The video has been around since then. Is Musk's argument that this was a long con to sue Tesla seven years later?
And then you have the concept that all it would take is alternate footage of same statement, B-roll, etc., to add to the "what, were all these deep fakes, too?"
How would you establish this video has been around since 2016? I could easily upload the same video and put a caption under it of "filmed shortly after the birth of Elon Musk" on it.
And could you also fake news articles referencing it all the way back to then?
Or the Internet Archive?
At some point you recognize that "reasonable doubt" doesn't mean "insurmountable evidence to disprove even the worse conspiracy theorist/asshole wrong".
> I could easily upload the same video and put a caption under it of "filmed shortly after the birth of Elon Musk" on it.
Except it's on YouTube, not a website under their control, which means you now need to argue - with convincing evidence - that YouTube is in on a conspiracy to defraud the court to get at Musk.
> The judge's ruling here is "ok then if you want me to think it's fake, tell me that under oath."
Which seems quite reasonable to me. Tbh, reading the initial article, I almost imagine Musk discussing his case with his lawyers, and saying something off the cuft like "can't we just argue that this was a deep-fake", and forcing them to run with it.
...because they're constantly being raked over the coals for being "biased", "liberal", and (now, conveniently thanks to Musk in large part): "state-controlled" perhaps?
If news events depict one side of the political spectrum unfavorably, "balanced" reporting plays directly into the agenda of that side. Truthful reporting of the news shouldn't always mean "all sides have valid arguments," and yet that is what our current media climate demands.
"Tesla lawyers argue Musk cannot be held accountable for public statements [about car safety] because anyone could have made a deepfake of Musk, so no public statements can be taken seriously." Court laughs.
This doesn’t seem unreasonable. Isn’t this already the case for politicians in the US? (Not related to deepfakes, just any public statement). Wouldn’t any legal matters defer to written contracts and TOS?
Indeed. Given the current state of affairs, I think if anyone were to be a victim of a deepfake smear campaing, Musk is easily in top 10 of the most likely candidates.
Then let Musk hire an expert who can act as a witness. Let the prosecution do the same. AKA - how technical concerns like this are already argued in court.
Photoshop, videos with doubles, and other such impersonation issues have been a thing for a long time, and the courts have found ways to deal with them.
It's very unreasonable when the actual person makes the actual statement. You know how people keep saying contracts aren't worth the paper they're written on? You know what's one of the things making your contract and TOS practically worthless? Your public statements.
I'm confused how courts will deal with this in the future. We are just entering the age of deep fake voices, convincing ones at least. And there are bound to be plenty of deep fakes that sound like the real person but aren't.
Are we now going to be held liable for deepfakes we didn't make?
I'm surprised people aren't using this more for divorces and what not. Create a deepfake of your wife and then say "Here's proof my wife was cheating"
Videos are easy enough to spot as fakes for now, but I've heard really convincing AI audio fakes. Audio only recordings might not be usable as evidence without additional confirmation of authenticity.
Are audio/video recordings I take with my phone going to be useless in a decade because deepfakes are virtually indistinguishable and there is no "additional confirmation"?
That is a good question, because there are two trends at play here. On the one hand, you have ever improving ability to create deepfake images, videos or audio recordings. On the other hand, AI-based retouching of photos and videos is now pretty much standard on mobile, some of it done by default and hidden from you in software. It wouldn't surprise me if the same was case with voice too.
Point being, it'll take more than just looking at the source image/video/audio for signs of AI manipulation, because soon enough, all recordings will be AI-manipulated before even hitting persistent storage.
In other words: everything will be deepfake by default, and the issue will turn into telling which deepfakes were done to lie about substantial aspects of reality pertinent to the case, and which ones were made to enhance some non-relevant qualities (aesthetics, compressibility, etc.).
Lots of ways to sign photos and videos today that should make them fully verifiable in the future (assuming public/private key cryptography remains secure). For stuff that isn't signed, it's not like good old detective work can't verify in a lot of cases. Phone records, eye witnesses, additional unpublished recordings, etc.
Also being verified really only matters when it comes to "truth" as defined by law. It's not like your family won't enjoy your videos of Christmas '23.
Additional confirmation could simply be multiple sources reporting the same thing. Once is circumstantial, but if 10 people record something from different perspectives then it is highly likely to be real.
This isn't a new problem with video, but a new twist on an old problem: manufactured evidence.
One way that courts deal with it is enhanced scrutiny on disputed evidence. Like, if you show up to a court with a document that you claimed someone signed, and they claim they didn't sign it, then the court can keep digging until they figure out who is lying. They can look for witnesses. They can subpoena records from internet companies that may have been involved in a digital transmission of that document. And so on. It's still possibly to lie to courts and manufacture evidence, but most people are not going to cover their tracks well enough to do so without getting caught if the full force the court's scrutiny is turned on them.
And the court finding out you lied to it is veryvery bad for whatever case you were trying to make.
Everything will be considered fake unless it is proven authentic.
How do we prove authentic? Same way we prove that a website we visit is authentic.
Every recording made will be signed/hashed and be able to validate of the public key of the recording device (every phone will have a key). White House, News Orgs, etc. all will use cameras with public keys so everyone can validate.
As soon as only one pixel or frequency changes, it will be invalid/fake.
In the end, all will be even more trustworthy than today (there will be some kind of signed/verification label on every video/audio/media that is original/valid).
I imagine the guy above is thinking any device can produce a hashed video, but I think we'll need the notion of cryptographically secure cameras or something with write once memory that has to be swapped out once in a while. The alternative being cameras that upload the hash of every clip into a server to provide non repudiation.
The idea is that the blockchain tracks the date the video was entered onto the blockchain verifiably. Note that this isn't the same as the date the images were taken, but it confirms that the recording is at least X days old.
But then the metadata, e.g. gps coords, time of the recording, etc. of the recorded video will not match with what is displayed. So it would be an obvious fake.
You know, contracts are just pieces of paper. They can easily be faked. Yet try going on court and say that a contract with your signature is invalid because it could be fake. Your argument will be rejected, because it doesn't matter if it could be faked, it matters if you are alleging that it was faked.
So the same thing should be the standard for video. Someone shows a video of you doing something. Either say under penalty of perjury, that you did not do that, and then the authenticity of the video can be looked into, or the video will be accepted as authentic. You don't get a preemptive "this could be fake, so lets ignore it".
So the way the lawyers are approaching this evidence is their usual cavalier approach to truth. They are not attempting to establish “my client never said that,” but just trying to poke holes, “this video of my client saying that is not sufficient proof that my client said that.”
There is for example no statement under oath from Elon, “that video was a fake, that absolutely never happened.” Even without that, there is no paper trail of “my client maintained a very consistent messaging that this was not fully autonomous driving and you need to always be vigilant... this isolated clip is so strange that we need to seriously entertain the idea that it was deepfaked.” That sort of stuff.
Regarding the faithful wife accused of cheating, I mean, there are lots of valid ways to investigate. First questions are “where did this video come from, where do you believe it was filmed, why do you believe it is your wife rather than someone who happens to look a lot like her,” etc. ... Normally a PI would be claiming “I followed her to so-and-so motel, I took this video legally from outside” and even then you might get objections from a decent divorce lawyer, “this is a two-party consent state and my client didn't consent, this is unlawful surveillance,” yadda yadda yadda. But for a deepfake you have all those objections and more.
Look up “laying the groundwork” for “documentary evidence” and you will find that this fails to show chain of custody unless you can get that “guy saying he is sleeping with her” in the courtroom.
It is 100% not an issue. The human eye and ears can’t tell if some video or audio is fake, but you cannot hide from the machine. Deep technical analysis will easily reveal something is manufactured, as easily as it can tell an image is photoshopped.
And God help you if they discover you manufactured evidence. Manufacturing evidence raises the stakes considerably. It makes any crime instantly more heinous, proves you deliberately had bad intention. You might go in thinking you’ll win easily with manufacturerd evidence and come out with greater felonies and complete eradication of your reputation.
You think you might be so fucking smart using off the shelf components to make yourself a deep fake, rest assured there are always people smarter than you out there. Intellects vast and deep that will easily pierce through your lies and destroy you. Do not attempt to provoke them. They will relish in picking your remains apart.
> I'm surprised people aren't using this more for divorces and what not. Create a deepfake of your wife and then say "Here's proof my wife was cheating"
No fault divorce makes it kind of pointless to use in divorces.
It's not a new problem. There was an HN article I was reading about someone facing police audio tape evidence and their defence lawyer said it must have been edited to make their client look guilty. The prosecution proved the recording was continuous and not edited from the pattern of mains hum noise on the tape. There'll probably be some clever technical methods to differentiate deep fakes from original recording along soon. Then there's just simply the weight of other evidence that may be present e.g. whether witnesses agree they said the statement etc.
I don't think the same will apply here. Even original record methods are dubious but for deepfakes it's just a matter of time to when it's completely indistinguishable from the real thing
Maybe we could do a blind test of the courts, sending them real and fake videos, and check which percentage of the videos are they really able to verify.
When courts still use lie detector tests and first person accounts as evidence (both of which have been proven to be close to useless), this will take a while for an actual solution to come by.
So-called "lie detectors" are based on pseudoscience. They are just as much "investigative tool" as a kazoo can be used for drilling a hole in concrete.
The way I've seen this work is that the Judge throws that in with 17 other bullshit arguments.
And unless you can take a thorough and comprehensive swipe at all eighteen, which takes about 160 pages of filings, then your odds of even being heard on appeal are small. Something like 90% of appeals that do get heard lose anyway.
This all costs a lot of money when lawyers cost $150 - $700 per hour. Expert testimony costs about the same. Appeals end up costing from $10k to $50k.
And how demoralized are you, to hear your judge saying things like that against you. US judges have little fear of accountability. The statistics there are abysmal.
No large media outlet will comment on such a decision. Few higher courts would ever inflict any punishment on the judge. To get punished as a judge you need to either provably ruin hundreds of lives , like in the kids-for-cash scandal [0], or one very wealthy and connected life.
OJ's bloody glove is probably the most famous example.
Bill Cosby's case was also very famous, and recent. Only one victim was allowed to testify - inexplicable.
The very recent Donziger case was pretty shocking; some major evidence was ignored there, along with established procedures.
Then you have things going on at even higher levels, like the Supreme Court deciding not to investigate itself, or the CFR blocking witnesses that would have said the WMD story was baloney.
Those were cases where the entire world was watching. Imagine what goes on where it's only a million or two at stake.
He certainly was railroaded. He had plenty of ironclad evidence and precedent on his side, and even a little media attention. It didn't matter.
There wasn't even a hint of repercussions (not even mild official rebuke) for the people that stitched Donziger up via the Justice system (Kaplan, Preska, Chevron, etc).
About Musk case, couldn't you simply gather enough people who were present and ask if this recording matches their recollection of what he said on that day? Just like most witness testimonies work.
I tell my lawyer friends who are early in their careers that they should build a practice in this area. It's going to be HUGE.
My personal opinion is that in a legal court setting that the burden of proof of an audio or video recording being real should be with the party presenting the evidence. Court standards for provenance need to be met before any recordings can even be shown to a jury.
Couldn't the original copy of a video be sha256-hashed by, say, the recording equipment firmware itself, and then possibly time/date-proofed somehow (such as on, sigh, a blockchain)?
Is there a need for a service like this? (I've been looking for something interesting to build.)
The general principle is called “chain of custody” [1] and it seems odd that it wasn’t mentioned in the article. Not a lawyer, but it normally doesn’t seem that rigorous? Someone testifies about how the video was recorded and stored, starting from the camera, to show that it’s unlikely to have been tampered with.
So it seems like if it comes from a security camera and you can say how it was recorded then that’s good, but if it’s video found on social media and you don’t know who recorded it then that should count against it.
I think before building anything, it might make sense to see what security cameras already do?
Sony, Nikon, and Canon already offer this for select "pro" cameras, specifically for law enforcement use.
Having unforgeable signatures on every image-capture device just in case some output file could be submitted as evidence in a civil case would be a privacy and safety _nightmare_, OTOH.
Nothing new in Indonesia, a supposedly pluralistic country where the parliament opens with Muslim prayer anyway. If a sex tape of yours get stolen and published, they'll drag you to court for "making pornography". When this happens to someone famous, their first defense is "It's fake!".
Essentially all evidence in court is assumed to be inadmissible (potentially fake) until a "foundation" is laid by the party who wishes to use it as an exhibit at trial.
I can't just turn up with a random photo or video and ask the court to accept it.
Generally this is done by having the person that created the media swear under oath. In the case of a photograph you would have the photographer on the stand and ask them to state that the photo they are shown is the one they took and that it hasn't been edited since they took it.
The (toothless) threat of perjury charges is what is supposed to keep this system working and our trials fair.
If you want to know more about it, post below. I have a fair bit of experience in this.
What is the process like in practice, and what are its weaknesses? How do you think this process will have to change to account for the ease of fabricating digital evidence?
OK, first thing to remember is that in most trials the lawyers for both parties probably know each other and the judge professionally, and often personally, depending a lot on the size of the courthouse/venue (technical term for the geographical area the court hears cases from). This is important because all three sides, plaintiff, defendant, and court (judge) want to make their life as simple as possible. In practice this means that before trial the judge will ask everyone to sort out the foundation before trial so that things can go quickly and in the case of a jury trial the jury's precious time isn't wasted on technical matters (as generally the jury isn't allowed to have an opinion on whether a piece of evidence is real - that is a matter solely for a judge to rule on).
What actually happens though, for the most part, is that none of the lawyers want to waste the time of a photographer to come into court to spend two seconds to say "yes" when asked if these are the photos they took. (Getting any witness into court is a nightmare). So the parties agree to "stipulate" the foundation of exhibits. So if the prosecutor asks to introduce a crime scene photo the defense lawyer will accept a stipulation that the photographer would swear to the images. This might buy the defense lawyer some brownie points as it saves the prosecution a lot of grunt work. The judge will also look on it favorably too, so more points there.
What is sad is that a lot of cases can be won on foundational grounds if lawyers stop stipulating to make things easier. There are many ways for foundation to fail and lawyers should require foundation to be proven on all but the most infallible cases.
One big issue coming through courts is child pornography. In many jurisdictions CSAM is legal if no real minors are involved in its production. To prove an image contains a real minor you either need the photographer (!) to vouch for it, or someone else that can verify the contents (e.g. the victim).
There are many other technical pieces to foundation that I could spend hours on. For instance, recorded security cam footage has no "photographer" so it is validated by swearing all the equipment was working fine.
The actual court Rules of Evidence for this process are really tiny, and practically all of the determinations are actually made based not on the wording of the rules but from the millions of pages of opinions of appeal court judges (case law) that has been written on this subject over the last few hundred years.
Deepfake detection will always be imperfect. The only solution the epistemological challenge from deepfakes is to securely authenticate legitimate media.
The Coalition for Content Provenance and Authenticity (C2PA -- c2pa.org) is an emerging standard to apply cryptographic signatures to media starting in the camera through the editing phase to distribution and consumption. Unfortunately C2PA does not yet have widespread adoption and we are facing a very challenging few years where we will not reliably be able to distinguish real and fake media.
I believe Nikon and Canon both (Canon certainly) absolutely have had this available for at least a decade or more, with hardware modules for their cameras that were able to cryptographically sign raw images to attest that that signed data was direct from the sensor. Was a huge concern once digital cameras became sophisticated enough to use in law enforcement.
> Suddenly there's no more reality," he said. "And that is really worrisome because I don't know how we reason about the world.
That seems a bit hyperbole. Sure, it's a regression compared to the state of five years ago. But in the bigger picture, it's only been a bit over a decade that recording devices have become omnipresent, maybe 50 years since retail-grade hardware has existed at all. It's not like humanity has no experience in dealing with situations where there's no (reliable) video evidence. For the longest time, hearsay was all we had. And it's not like we couldn't 'reason about the world' back then.
> In 2016, Elon Musk went on stage at a tech conference outside Los Angeles and made a bold statement about the self-driving capability of Teslas.
Apart from the video, wouldn't there have been lots of witnesses who where at the conference in person and saw him make that statement?
Even apart from the "burden of proof" debate, it seems kinda dumb to try and set this precedent in a case where the prosecution could actually provide proof of the video being genuine relatively easily.
(Unless you want to assume that Musk was playing 8D chess here and intentionally introducing the "everything is fake unless proven otherwise" defense in one of the rare situations where it could actually be disproven, to make it appear less devastating to the judge than it actually is. If that was the plan, the judge wasn't having it though.)
If you watch the video, he talks about not self driving in general, but self driving on highways that have barriers. That part, he says, is where autopilot is already a safer driver than a person. It’s around the 1h19m33s mark of the Recode video that’s linked in the article.
198 comments
[ 3.8 ms ] story [ 257 ms ] threadhttps://www.law.cornell.edu/wex/officer_of_the_court
No, its not.
> and there is no penalty or sanction for it
Yes, there is.
> - especially not against the lawyers themselves.
There is a penalty especially against the lawyers. Heck, lawyers have been suspended from practice for lying to courts outside of their capacity as lawyers.
This is tantamount to, but NOT equivalent to, "lying" in my book. Some of the most famous defenses in history, such as Cochran's "if it doesn't fit, you must acquit" https://www.youtube.com/watch?v=NH-VuP_5cA4 were fundamentally, fallacious appeals in nature.
Hugely disappointing to me. Would love for a lawyer to chime in on this.
If I could come up with any defense of this practice at all, I'd say that if you took my reasoning to its logical extreme, then the persuasive personality of an attorney themselves could be considered a fallacy, and that would be an unreasonable expectation to satisfy the elimination of. You could also argue that if 2 opposing lawyers are both permitted to make fallacious appeals that wouldn't be called out by judges or jury, then in theory they'd mostly cancel each other out. Maybe.
https://www.law.cornell.edu/wex/officer_of_the_court
It definitely carries penalties for the attorney. Each US state has passed 'rules of professional conduct' that govern rules that apply to barred attorneys. Lying to a tribunal, or lying to your client, is illegal and you can be disbarred and fined for it.
Haven't read this case closely but it sounds to me like the attorneys said "it's possible these are deep fakes" and not "these are deepfakes." The former seems like a reasonable argument to make, but according to the judge, a flawed one.
That said, at the most important moments, typically lawyers will focus on the 3-or-so 'best' (in their view) defenses and focus on those. Then if you lose, you can appeal and say the judge didn't consider your 15th-best defense closely enough :)
IANAL, but this sounds appropriate to me? Couldn't all forms evidence be hit with various types of "this might be fake" claims? So I think it's fair that the court assert that you must claim "this is fake", then be willing to battle it out with the opposing side. Otherwise it's not a defense, it's just empty words, isn't it?
> Tesla could not admit or deny the authenticity of video and audio recordings purportedly containing statements by Mr. Musk
https://arstechnica.com/tech-policy/2023/04/judge-slams-tesl...
Federal courts have rule 11, where an attorney can essentially call out the opposing counsel after some cure period and ask the court to impose sanctions on the attorney that's getting out of line. State courts have some equivalent.
In practice, most lawyers don't lie to the court because their career is dependent on maintaining a working relationship with the courts in which they appear. Even within the same case, if the presiding judge has good reason to believe you're a liar, every subsequent motion or argument you make is going to be viewed with skepticism.
> "Their position is that because Mr. Musk is famous and might be more of a target for deep fakes, his public statements are immune," she wrote. "In other words, Mr. Musk, and others in his position, can simply say whatever they like in the public domain, then hide behind the potential for their recorded statements being a deep fake to avoid taking ownership of what they did actually say and do. The Court is unwilling to set such a precedent by condoning Tesla's approach here."
It's no more "guilty until proven innocent" than a signed confession means you're guilty unless you prove it fake; that's the whole point of a trial.0
For a Jury i think it would come down to how convincingly the defense argued that the video is a deepfake. Just saying the words "it's a deepfake" hypothetically wouldn't/shouldn't be enough to convince a "reasonable person" to believe all video evidence is now fake without concrete evidence (if any is even possible) from the prosecution.
Disclaimer: I'm quite unfamiliar with the nuance of the legal system. But i keep using the words "reasonable person" because this was hammered into us in the Jury. We were given a small book of legal definitions related to the criminal case and just about every page framed each definition around a "reasonable person". This seemed to be how the state defined any nuance between the English language and strict logic.
This would, i imagine, help mitigate a hypothetical defense of "all evidence is faked" or w/e.
I think you're thinking of reasonable doubt?
Arguing that a video randomly was deepfaked would go a lot further if you could show other evidence that such was happening.
For additional context, the criminal case i was part of the Jury for was handed a book of definitions for the given case. Things like Assault had qualifiers of Self Defense and Stand Your Ground. Each of these little nuance-y definitions included very implicit definitions of say, the force you use in response to assault for self defense or stand your ground. This responsive force was instructed to be judged against something like "Force that a reasonable person would not believe to be in excess based on the attack".
Ie "reasonable" was only able to be decided by a hypothetical "reasonable person". Whether something is reasonable or not was only applicable in the context of the person who is deciding if it's reasonable, also themselves being a reasonable person.
Just about every definition page had an explanation quite similar to this. I think due to the fact that human definitions are very difficult to define. Did you hit them too hard? Well we don't have pounds per square inch of your delivery, or w/e concrete definition could use, so the legal definitions were bounded by this hypothetical "reasonable person".
It was interesting.
edit: I should add, in case it's not clear, that we the Jury were instructed by the judge on the legal definitions. That was the book. We were to view the case through the lens of these legal definitions with the evidence presented. What specifically we were deciding was very important and meticulously managed by the Judge.
The "reasonable doubt" comes into play on the facts.
(The system actually works pretty well, and has been developed over long periods of time.)
So it’s entirely possible for a ton of evidence to come in that none of it meets what you’d consider the “no reasonable doubt” standard and for a jury to consider all that suspicious evidence to be true enough that they can convict someone under the “no reasonable doubt” standard.
The fact that something is presented as evidence doesn’t mandate the jury to accept the evidence as dispositive.
It means the same thing that's always been true remains true. If a prosecutor (or one side in a civil case) presents fake evidence that is convincing to the court, the other side will have to prove the evidence to have been faked. Adding machine learning to the equation and calling the fakes "deep" doesn't change that basic reality.
Consider everyone who has ever had drugs/weapons planted on them by the cops. As a society we assume most of the people claiming it happened are lying, but some of them are telling the truth and some of that subset actually manage to prove it (usually because the cops managed to record themselves doing it). If the authority figures trusted by the courts intentionally present false evidence then you are stuck not having to prove your innocence but having to prove the evidence to be false or at least untrustworthy.
It's really hard to predict where the technology is headed with regards to both generating and proving the non-authenticity of deepfakes, but it is concerning that this could open up people without money to pay for defense to be ruined by people with money to pay for deepfakes.
The defense's first step is to say "that particular video is a deep fake", under penalty of perjury.
Musk's "well it could be" doesn't cut it.
It's also not clear if experts always can tell if something is a fake or not, if it was created by an algorithm which is less than a year old or doesn't happen to have a lot of the artifacts that these algorithms sometimes introduce.
The courts will have to deal with it one way or another, and right now expert witnesses is the tool they have, and the tool defendants will need to use.
Why not?
If you go into court claiming you have a contract with someone but you have nothing notarized or even signed, you're the one who has to provide evidence of your claim. A video is no longer sufficient to do that because any plaintiff or internet troll now has the ability to convincingly fabricate one. It's no more proof of anything than the plaintiff making an unsubstantiated claim.
> The courts will have to deal with it one way or another, and right now expert witnesses is the tool they have, and the tool defendants will need to use.
Why should it be the defendants who have to pay for the expert instead of the plaintiffs?
> Why should it be the defendants who have to pay for the expert instead of the plaintiffs?
TL;DR: that's how our legal system works in the US, for better or worse.
EDIT: You probably don't have to attack the video as a deepfake directly, either. "That's not me, I wasn't at X, I was at Y, and here are receipts and witnesses" are highly likely to work too. Basically, an alibi.
Suppose the video isn't of an old statement made by the defendant which they don't remember making, it's a video of one of their products catching fire sourced from an anonymous YouTube account. They have no way to know if that video is fake or not, so on what basis could they claim to know for sure that it is?
But because it could very plausibly be fake, they should be able to demand proof that it isn't from the party trying to introduce it as evidence. For example, by having it authenticated by the person who created it, whose credibility could then be questioned etc.
> "That's not me, I wasn't at X, I was at Y, and here are receipts and witnesses" are highly likely to work too. Basically, an alibi.
An alibi is the defense proving that they couldn't have done it. The burden of proof is supposed to be on the plaintiff. Creating deepfakes is so easy that a plaintiff in possession of one by itself proves nothing.
As for the rest - you'll have more luck discussing the specifics with your lawyer at this point. Good luck and I hope you find what you need.
This sounds ridiculous since we lived most of our lives in the video-is-real era that has sort of been coming to an end. I recall the Kyle Rittenhouse trial had an interesting take on this where the defense asked to throw out upscaled video which contrary to every police procedural TV show is a lot like just fabricating data. With the amount of post processing and AI a cell phone camera does these days you could argue everything they pump out is doctored and fake by default because it kind of is. And most video is stored with lossy algorithms even without the post processing.
From the federal rules of evidence:
> (a) In General. To satisfy the requirement of authenticating or identifying an item of evidence, the proponent must produce evidence sufficient to support a finding that the item is what the proponent claims it is.
https://www.rulesofevidence.org/article-ix/rule-901/
Specific to recordings
> An original writing, recording, or photograph is required in order to prove its content unless these rules or a federal statute provides otherwise.
https://www.rulesofevidence.org/article-x/rule-1002/
Unfortunately, the rules of evidence are pretty vague about what is actually required to prove that a video is admitable as evidence; but the burden to do so is absolutely on the party introducing it.
If you have that (absent other possible evidentuary concerns), it is up to the trier of fact (e.g. jury) to determine if how much they believe the evidence.
The problem here is the prosecution cannot meet even that low bar, and instead is relying on the various exceptions to the requirement.
Rule 902 lists evidence that is "self authenticating", and videos are not on it.
https://www.rulesofevidence.org/article-ix/rule-902/
At that point the reliability of the evidence may of course be called into question, as with any evidence or witness. The question mark would be whether it was convincing.
As such, the question of deep-fakes should be relatively irrelevant to (criminal) courts. If a witness claims that a particular video was filmed on a particular day in a particular way, that is as much evidence as you need. Whether that witness is lying remains to be proven during the course of the trial.
Things may be different in civil court, where the standards for evidence are more lax I believe.
Prosecution has a freer reign during deposition and can play a video from Youtube[1] and ask the subject of the video if they were at the location at a specific date and time, and if the words are their own.
1. With the courts permission.
For example, lets say someone accused of a violent crime and the prosecutor wants it upgraded to a hate crime. Now lets the accused did not post a life of hate on facebook. But instead there are just a few audio recordings of the accused saying "I hate X people, I want to kill them all" brought to the court by someone else and not found in the accused possession.
As time progresses it will likely become more difficult to prove that the audio is real. Deepfakes of audio are convincingly good these days. Soon the recordings of politicians saying terrible things will be met with a default reply of "it's a deepfake", and it's likely a bunch of them will be.
Future is going to be messy, yo.
If any evidence was automatically disqualified on the basis that it merely could be fake, there could never be any successful prosecution.
The defense is still welcome, of course, to make an argument and present evidence that any video or other evidence provided by the prosecution is fake.
Substantiating your evidence has to happen via a pretty specific process, and there is an entire process for entering stuff into evidence with rules regarding chain of custody, validation, etc.
When it comes to evidence, it's not a specific claim against anyone, it's an item that is being introduced to support a claim. That should be validated and checked before it even gets to the court room. If either counsel believes it to be fraudulent, they need to review it and make the claim.
From the article, this was not really the case; instead, a video of a conference with many attendees and participants showed Musk making the claim in question.[0] Tesla's (or Musk's? Not sure) lawyers made a very weak statement:
"[Musk], like many public figures, is the subject of many ‘deepfake’ videos and audio recordings that purport to show him saying and doing things he never actually said or did,” Tesla said"
This is not really an argument towards anything about the video; it doesn't outright say it's a deepfake (likely because it's not), it doesn't provide any evidence to a claim that the video in question is a deepfake, it does nothing to establish who would have produced this deepfake, it doesn't question the YouTube channel owners, Recode, on if they are ready to stand by their statement, etc.
In this instance, the lawyers defending Tesla/Musk are trying to muddy the waters; they aren't bold enough to say Recode made a deepfake, they aren't trying hard to position that the video _is_ a deepfake, they're just trying to sow doubt into the minds of the court about whether to consider normal evidence like recordings or screenshots as valid given that Musk is a public figure. (not sure if there's a jury, but if there is, they want to confuse the jury also)
The judge was right in my opinion to slap the defense counsel on this, as it's transparent and pithy. It was a bad-faith statement to make for a legal team in my opinion as they weren't trying to _say it was a fake_, they just wanted to sow doubt without any proof.
It's even a bit hopeful in my opinion as to me it says the court is holding Musk and the businesses he associates with accountable for Musk's actions and statements; Musk has tried this before when he "joked" about buying Twitter and the courts didn't take on any of the arguments they put up. As I see it, it's another instance of Musk and counsel representing his interests trying desperately to escape culpability.
0 - https://www.youtube.com/watch?v=wsixsRI-Sz4&t=4765s
Edit: Wrong link on first version. I was trying to see if any other channels had coverage of the Code Conference 2016, but looks like it's just Recode.
There should be higher degree of burden on the defense than claiming "fake news" and waving your hand.
How believable the emails are will vary. Do you still have the headers? Is there a valid DKIM signature in those headers? Can you bring the people CCed on the email as witnesses? etc.
For an old email, it would be up to the opposing counsel to argue that the person who made it made it for the purpose of fabricating evidence however many years ago.
The judge is very much for the prosecution deposing Musk to ask him - under oath - if he can corroborate or dispute what's in the video. The defense was asking the videos to be thrown out entirely because they could be fake.
I’m not a lawyer, but don’t courts have a way of weighting the trustworthiness of a piece of evidence?
In Mr. Musks case, there are many witnesses who observed him making the statements that are shown in the video.
however the courts have been lack of late allowing police to download social media and YT videos and attest that "the video is a true and accurate representation of what I saw on social media post" because they often can not find, or can not secure the person that actually took the video (or the video was taken by the defendant who can not be made to testify against themselves)
It is very dangerous when we start lessing the requirements for foundation in the law, and the courts have been doing that on many fronts when it comes to "forensic science" (see bite mark evidence as another example of this)
From a linked article from the one posted here, https://www.theguardian.com/technology/2023/apr/27/elon-musk... :
> The attorneys for Huang’s family sought to depose Musk regarding recorded statements from 2016 in which he allegedly said: “A Model S and Model X, at this point, can drive autonomously with greater safety than a person. Right now.”
> Tesla, however, opposed the request in court filings, arguing that Musk, the Tesla CEO, cannot recall details about the statement and questioning the authenticity of the recording.
The judge's ruling here is "ok then if you want me to think it's fake, tell me that under oath."
It's doubtful Musk would be willing to claim this one was fake - there are a lot of witnesses that could be dug up. He could do the weaker "I don't remember" song and dance, but he's trying to just avoid the issue entirely.
How do you propose to prove if he actually remembers or not?
> It's doubtful Musk would be willing to claim this one was fake - there are a lot of witnesses that could be dug up.
But then the prosecution doesn't need the video because they would have all of these witness, so why should it be admissible if it can't be authenticated?
He doesn't have to remember it for it to be good evidence. If he wants to say "I don't think I would've said that", go for it.
> But then the prosecution doesn't need the video because they would have all of these witness, so why should it be admissible if it can't be authenticated?
The video is perfectly good evidence (and its age and provenance make it pretty reputable). If the defense contests its veracity, that's when you start going to witnesses.
He doesn't have to remember it for it to be bad evidence either.
> If he wants to say "I don't think I would've said that", go for it.
But then what are you trying to hold him to? A subjective opinion?
> The video is perfectly good evidence (and its age and provenance make it pretty reputable). If the defense contests its veracity, that's when you start going to witnesses.
The defense will always contest its veracity if the burden is on the prosecution to prove it and they haven't. The point is that they should have to prove it.
Sometimes they'll be able to.
He needs to make a claim about the evidence in question if he thinks it's fake. (He doesn't. He knows it's a genuine video.) Again, a vague assertion that "sometimes people make deepfakes about me" doesn't cut it.
> The defense will always contest its veracity if the burden is on the prosecution to prove it and they haven't.
The defense is highly unlikely to make the claim that a seven year old video on a prominent channel with top industry reporters didn't happen. Plenty of pieces of evidence get entered into the court's records without being contested.
Why? What if he thinks it could be fake, and doesn't know? Why should that reduce the other party's burden?
> The defense is highly unlikely to make the claim that a seven year old video on a prominent channel with top industry reporters didn't happen.
But it's not the defense claiming that it didn't happen, it's the plaintiff claiming that it did. They should have to prove their claims.
And sometimes they'll be able to.
> Plenty of pieces of evidence get entered into the court's records without being contested.
They don't get contested when the defense doesn't want to aggravate the judge by demanding proof of something they know the other party has, even if they have a right to. That's a different question than whether the other party should have the burden of proof in the cases where they might not be able to meet it.
He will be asked to justify that belief, and that justification will be weighed against the provenance of the video. That he's not saying that is telling; claiming the video is faked would likely hurt his credibility given the number of times he's provably made statements of this kind.
> They should have to prove their claims.
They have submitted it as evidence. The jury weighs if it's credible. If Musk chooses not to rebut/contest it, that's on him, not the plaintiff.
Why? His lawyers could provide the same justification: Convincing deepfakes are easily produced at low cost by adversarial plaintiffs and internet trolls.
> That he's not saying that is telling
People are imperfect and if you misspeak under oath there are consequences. Testifying at all subjects you to cross-examination, which is time consuming and subjects you to lawyer tricks by the other party. Not wanting to do it is the default.
> They have submitted it as evidence. The jury weighs if it's credible.
First the judge decides if it's even admissible. There are some forms of evidence so weak that they only serve to prejudice the jury, and unauthenticated video is pretty much there now.
Demanding that the introducing party authenticate it should be table stakes at this point.
Cool. Video evidence, written evidence, witness testimony, and essentially everything else is now off the table. Congratulations; you've gotten rid of the entire trial court system.
> Testifying at all subjects you to cross-examination, which is time consuming and subjects you to lawyer tricks by the other party. Not wanting to do it is the default.
Yes. If you aren't willing to argue against a piece of evidence in a convincing fashion, you have to live with that decision. The jury gets to draw inference from "huh, they didn't say anything in response to that..."
> Demanding that the introducing party authenticate it should be table stakes at this point.
They'll introduce it with someone describing it something like this:
"This is a seven year old video hosted publicly on YouTube, on the channel of a long-standing reputable organization, of a live event hosted by well-known reporters, which received contemporary news coverage. The claims in the video made by Musk are of a similar nature to others he has demonstrably made."
That's sufficient. Musk's lawyers are free to object, if they've got a good argument against all of that. They're also welcome to try and convince the jury that Musk never made the claims. I suspect the plaintiffs will be able to find plenty of corroborating evidence should it come to that.
You're going to create a deepfake of witness testimony in front of a live jury? How does that work?
And it's not that you can't submit a video, it's that if you want to, you should have to go find the person who took the video and have them testify that what it shows is what actually happened. And have them be identified, so their credibility can be called into question, e.g. if it turns out they're a long-time antagonist of the defendant.
> Yes. If you aren't willing to argue against a piece of evidence in a convincing fashion, you have to live with that decision. The jury gets to draw inference from "huh, they didn't say anything in response to that..."
But they did have something to say in response to that: Deepfakes are cheap and easy to create. That's a general argument, not a claim about the facts of the case, so it doesn't require testimony to substantiate.
> "This is a seven year old video hosted publicly on YouTube, on the channel of a long-standing reputable organization, of a live event hosted by well-known reporters, which received contemporary news coverage. The claims in the video made by Musk are of a similar nature to others he has demonstrably made."
None of which proves that the video is real. At best it's evidence that it's a possible fake which is seven years old and was published by an established organization.
What they could do is find the reporter who did the interview and bring them in to testify that what the video depicts is what actually happened, if they have a better memory of it than Musk. Which they may very well be able to do. But the point is that they should have to.
Witness testimony is trivially fabricated. It happens all the time in court. The solution is to rebut it, not ban it.
> But they did have something to say in response to that: Deepfakes are cheap and easy to create.
That's not a reasonable response, any more than "I saw him shoot the person" can't be rebutted with a vague, hand-wavy "well sometimes people lie" by itself.
> None of which proves that the video is real. At best it's evidence that it's a possible fake which is seven years old and was published by an established organization.
I don't think you're arguing in good faith.
Witness testimony is under oath. You can lie, but then you can go to jail. And you can be cross-examined, unlike a video. The risk of punishment reduces the incentive to lie and the interactivity increases the potential for it to be discovered.
Witnesses also have varying credibility. If the witness is the plaintiff, they have an obvious incentive to lie and juries know that. Whereas a video from an unauthenticated source is no more credible but fosters the intuition that you're seeing something with your own eyes.
> I don't think you're arguing in good faith.
I think I am.
I think we can agree that a random video from a random YouTube channel has no credibility. But if you're going to stake the value of your evidence on the credibility of the source, you should at least have to establish that it can actually be authenticated by one of these credible parties and not just e.g. some syndicated clip that got published on their channel.
Which, once again, you may be able to do. But you should still have the burden of actually doing it.
if all elmu has to say in response is "well maybe it's a deepfake, but I won't say if I think it is", obviously any jury would be hard pressed to suppress their pressing giggles before rejecting such a weak excuse
There's not a "prosecution." Did you read the article I linked? This is a civil lawsuit (which also means a different level of evidence required, ultimately) still in fairly early stages. The trial hasn't even started. "Judge Evette Pennypacker tentatively ordered a limited, three-hour deposition where Musk could be asked whether he actually made the statements on the recordings" - there's no ruling of fact yet about if the video is real, the Judge is just saying "this looks like a bullshit excuse to avoid having to talk about your past comments."
The plaintiff is trying to get Musk to give a deposition regarding his claims about self-driving. Musk is trying to not even have to give that testimony. This is exactly that "we need to figure out if this evidence is real or not" process that you say is needed! Musk is just trying to throw that out and say "nah we don't even need it, it could be fake, ya know." The judge seems to think that a 6+ year old video from a fairly well-known publication carries enough weight compared to a lazy "IDK, sometimes people deep fake me" statement to be worth asking Musk further questions. That seems reasonable to me.
> How do you propose to prove if he actually remembers or not?
Obviously that's impossible, but it's also irrelevant. If he says he doesn't remember, but you have a video with an uploaded-to-Youtube-timestamp from 2016, and contemporary coverage, then you say "well, since you don't remember for sure, we're gonna go with all the evidence that suggests you did say this."
Also, how good was deep faking in 2016? The video has been around since then. Is Musk's argument that this was a long con to sue Tesla seven years later?
And then you have the concept that all it would take is alternate footage of same statement, B-roll, etc., to add to the "what, were all these deep fakes, too?"
Or the Internet Archive?
At some point you recognize that "reasonable doubt" doesn't mean "insurmountable evidence to disprove even the worse conspiracy theorist/asshole wrong".
Except it's on YouTube, not a website under their control, which means you now need to argue - with convincing evidence - that YouTube is in on a conspiracy to defraud the court to get at Musk.
Which seems quite reasonable to me. Tbh, reading the initial article, I almost imagine Musk discussing his case with his lawyers, and saying something off the cuft like "can't we just argue that this was a deep-fake", and forcing them to run with it.
If news events depict one side of the political spectrum unfavorably, "balanced" reporting plays directly into the agenda of that side. Truthful reporting of the news shouldn't always mean "all sides have valid arguments," and yet that is what our current media climate demands.
He can't say "because I'm the target of deep fakes, you can't use any video evidence against me".
Photoshop, videos with doubles, and other such impersonation issues have been a thing for a long time, and the courts have found ways to deal with them.
What you say matters. A lot.
Are we now going to be held liable for deepfakes we didn't make?
I'm surprised people aren't using this more for divorces and what not. Create a deepfake of your wife and then say "Here's proof my wife was cheating"
Are audio/video recordings I take with my phone going to be useless in a decade because deepfakes are virtually indistinguishable and there is no "additional confirmation"?
Point being, it'll take more than just looking at the source image/video/audio for signs of AI manipulation, because soon enough, all recordings will be AI-manipulated before even hitting persistent storage.
In other words: everything will be deepfake by default, and the issue will turn into telling which deepfakes were done to lie about substantial aspects of reality pertinent to the case, and which ones were made to enhance some non-relevant qualities (aesthetics, compressibility, etc.).
Lots of ways to sign photos and videos today that should make them fully verifiable in the future (assuming public/private key cryptography remains secure). For stuff that isn't signed, it's not like good old detective work can't verify in a lot of cases. Phone records, eye witnesses, additional unpublished recordings, etc.
Also being verified really only matters when it comes to "truth" as defined by law. It's not like your family won't enjoy your videos of Christmas '23.
One way that courts deal with it is enhanced scrutiny on disputed evidence. Like, if you show up to a court with a document that you claimed someone signed, and they claim they didn't sign it, then the court can keep digging until they figure out who is lying. They can look for witnesses. They can subpoena records from internet companies that may have been involved in a digital transmission of that document. And so on. It's still possibly to lie to courts and manufacture evidence, but most people are not going to cover their tracks well enough to do so without getting caught if the full force the court's scrutiny is turned on them.
And the court finding out you lied to it is very very bad for whatever case you were trying to make.
good luck doing that without severely deep pockets
How do we prove authentic? Same way we prove that a website we visit is authentic. Every recording made will be signed/hashed and be able to validate of the public key of the recording device (every phone will have a key). White House, News Orgs, etc. all will use cameras with public keys so everyone can validate. As soon as only one pixel or frequency changes, it will be invalid/fake.
In the end, all will be even more trustworthy than today (there will be some kind of signed/verification label on every video/audio/media that is original/valid).
So the same thing should be the standard for video. Someone shows a video of you doing something. Either say under penalty of perjury, that you did not do that, and then the authenticity of the video can be looked into, or the video will be accepted as authentic. You don't get a preemptive "this could be fake, so lets ignore it".
I wonder if there is any traction in the industry towards making this happen?
There is for example no statement under oath from Elon, “that video was a fake, that absolutely never happened.” Even without that, there is no paper trail of “my client maintained a very consistent messaging that this was not fully autonomous driving and you need to always be vigilant... this isolated clip is so strange that we need to seriously entertain the idea that it was deepfaked.” That sort of stuff.
Regarding the faithful wife accused of cheating, I mean, there are lots of valid ways to investigate. First questions are “where did this video come from, where do you believe it was filmed, why do you believe it is your wife rather than someone who happens to look a lot like her,” etc. ... Normally a PI would be claiming “I followed her to so-and-so motel, I took this video legally from outside” and even then you might get objections from a decent divorce lawyer, “this is a two-party consent state and my client didn't consent, this is unlawful surveillance,” yadda yadda yadda. But for a deepfake you have all those objections and more.
And God help you if they discover you manufactured evidence. Manufacturing evidence raises the stakes considerably. It makes any crime instantly more heinous, proves you deliberately had bad intention. You might go in thinking you’ll win easily with manufacturerd evidence and come out with greater felonies and complete eradication of your reputation.
You think you might be so fucking smart using off the shelf components to make yourself a deep fake, rest assured there are always people smarter than you out there. Intellects vast and deep that will easily pierce through your lies and destroy you. Do not attempt to provoke them. They will relish in picking your remains apart.
This is a ridiculous comment
No fault divorce makes it kind of pointless to use in divorces.
Even when I uploaded the video to YouTube within 24 hours of the wreck the "judge" said I had years to fake it.
And unless you can take a thorough and comprehensive swipe at all eighteen, which takes about 160 pages of filings, then your odds of even being heard on appeal are small. Something like 90% of appeals that do get heard lose anyway.
This all costs a lot of money when lawyers cost $150 - $700 per hour. Expert testimony costs about the same. Appeals end up costing from $10k to $50k.
And how demoralized are you, to hear your judge saying things like that against you. US judges have little fear of accountability. The statistics there are abysmal.
No large media outlet will comment on such a decision. Few higher courts would ever inflict any punishment on the judge. To get punished as a judge you need to either provably ruin hundreds of lives , like in the kids-for-cash scandal [0], or one very wealthy and connected life.
> https://en.wikipedia.org/wiki/Kids_for_cash_scandal
Bill Cosby's case was also very famous, and recent. Only one victim was allowed to testify - inexplicable.
The very recent Donziger case was pretty shocking; some major evidence was ignored there, along with established procedures.
Then you have things going on at even higher levels, like the Supreme Court deciding not to investigate itself, or the CFR blocking witnesses that would have said the WMD story was baloney.
Those were cases where the entire world was watching. Imagine what goes on where it's only a million or two at stake.
There wasn't even a hint of repercussions (not even mild official rebuke) for the people that stitched Donziger up via the Justice system (Kaplan, Preska, Chevron, etc).
https://www.imdb.com/title/tt8201186/
My personal opinion is that in a legal court setting that the burden of proof of an audio or video recording being real should be with the party presenting the evidence. Court standards for provenance need to be met before any recordings can even be shown to a jury.
There are plenty of billionaires who stay far more out of the limelight than either of these two.
You want praise? You have to accept scorn.
Is there a need for a service like this? (I've been looking for something interesting to build.)
So it seems like if it comes from a security camera and you can say how it was recorded then that’s good, but if it’s video found on social media and you don’t know who recorded it then that should count against it.
I think before building anything, it might make sense to see what security cameras already do?
[1] https://en.m.wikipedia.org/wiki/Chain_of_custody
Having unforgeable signatures on every image-capture device just in case some output file could be submitted as evidence in a civil case would be a privacy and safety _nightmare_, OTOH.
I can't just turn up with a random photo or video and ask the court to accept it.
Generally this is done by having the person that created the media swear under oath. In the case of a photograph you would have the photographer on the stand and ask them to state that the photo they are shown is the one they took and that it hasn't been edited since they took it.
The (toothless) threat of perjury charges is what is supposed to keep this system working and our trials fair.
If you want to know more about it, post below. I have a fair bit of experience in this.
What is the process like in practice, and what are its weaknesses? How do you think this process will have to change to account for the ease of fabricating digital evidence?
What actually happens though, for the most part, is that none of the lawyers want to waste the time of a photographer to come into court to spend two seconds to say "yes" when asked if these are the photos they took. (Getting any witness into court is a nightmare). So the parties agree to "stipulate" the foundation of exhibits. So if the prosecutor asks to introduce a crime scene photo the defense lawyer will accept a stipulation that the photographer would swear to the images. This might buy the defense lawyer some brownie points as it saves the prosecution a lot of grunt work. The judge will also look on it favorably too, so more points there.
What is sad is that a lot of cases can be won on foundational grounds if lawyers stop stipulating to make things easier. There are many ways for foundation to fail and lawyers should require foundation to be proven on all but the most infallible cases.
One big issue coming through courts is child pornography. In many jurisdictions CSAM is legal if no real minors are involved in its production. To prove an image contains a real minor you either need the photographer (!) to vouch for it, or someone else that can verify the contents (e.g. the victim).
There are many other technical pieces to foundation that I could spend hours on. For instance, recorded security cam footage has no "photographer" so it is validated by swearing all the equipment was working fine.
The actual court Rules of Evidence for this process are really tiny, and practically all of the determinations are actually made based not on the wording of the rules but from the millions of pages of opinions of appeal court judges (case law) that has been written on this subject over the last few hundred years.
The Coalition for Content Provenance and Authenticity (C2PA -- c2pa.org) is an emerging standard to apply cryptographic signatures to media starting in the camera through the editing phase to distribution and consumption. Unfortunately C2PA does not yet have widespread adoption and we are facing a very challenging few years where we will not reliably be able to distinguish real and fake media.
That seems a bit hyperbole. Sure, it's a regression compared to the state of five years ago. But in the bigger picture, it's only been a bit over a decade that recording devices have become omnipresent, maybe 50 years since retail-grade hardware has existed at all. It's not like humanity has no experience in dealing with situations where there's no (reliable) video evidence. For the longest time, hearsay was all we had. And it's not like we couldn't 'reason about the world' back then.
Apart from the video, wouldn't there have been lots of witnesses who where at the conference in person and saw him make that statement?
Even apart from the "burden of proof" debate, it seems kinda dumb to try and set this precedent in a case where the prosecution could actually provide proof of the video being genuine relatively easily.
(Unless you want to assume that Musk was playing 8D chess here and intentionally introducing the "everything is fake unless proven otherwise" defense in one of the rare situations where it could actually be disproven, to make it appear less devastating to the judge than it actually is. If that was the plan, the judge wasn't having it though.)
If you watch the video, he talks about not self driving in general, but self driving on highways that have barriers. That part, he says, is where autopilot is already a safer driver than a person. It’s around the 1h19m33s mark of the Recode video that’s linked in the article.