39 comments

[ 3.1 ms ] story [ 95.5 ms ] thread
As she should -- it's a slam dunk case.

You don't get to use a celebrity in your ad if they don't agree to it. AI doesn't change that.

What about my code?
Do you make a living off your code's likeness? That is, is there a wholesomepotato device, like Duff's or Not John Carmack's fast inverse square root, that's immediately recognizable as having been created by you, and you get paid for its appearance on late night tv shows and in movies?
copilots whole job is to make code look like it functions... arguably, that's entirely what these systems do.

the AI isn't doing fundamentals in programming anymore than it's putting on makeup to look like an actress.

if you go out and find a woman whose shaped like her, get a makeup artist and a costume designer and put together a simulacrum ... that's also illegal right?

anyway, it's pretty difficult to judge one as being ethically dubious and the other as hand wavy concern trolling.

Yeah, that too depending on what you mean. Just don't get carried away claiming things like "They saw I have a head and now the people in the generated ads have heads too" or "my code showed how to use parts of the Box2D library and their code does too" are the same type of issue as "they used my public data to impersonate me" or "my code is being stolen". Maybe you want to argue the latter things are still problematic but no amount of pointing it out in situations like this achieves that.
You can use celebrities in your code as long as they're open source.
We are entitled to it.

The system cannot be compromised by a developer retracting their code from the machine. A developer cannot be allowed authority over the machine. Code is powerful, unlike ScarJo's likeness.

Knowledge workers are permanently forced to lose control of their own labour.

It's how it is.

if your code contributed to any AI project, it sure must suck because I rarely get code from any AI that runs on the first try. If it was verbatim, which is the only claim that is valid, shouldn't it work every time? How do you define the "likeness" of a codebase, and someone duplicating that at least stylistically, and how would that accusation be defended in court?
Really?

How do attack ads exist?

Seems impossible.

Attack ads reflect the actual behaviour of a person, actual footage etc. Image generation doesn't.
So the problem isn't that it's too real; it's that it's not real enough.
When the phrase whataboutism first became popular I really disliked it, but now that the idea is in my head I see things like this post constantly.

The law protects certain uses of things like copyright because the societal benefit outweighs any real cost, political opponents being able to use each others likeness in an ad is not nearly the same as a company getting the use of a celebrity without paying for it.

And please don't start with the "attack ads harm society", you chose to limit your description to attack ads when what you really meant was political ads (attacking or otherwise).

For the same reason the news exists.

CBS News is allowed to show footage of ScarJo speaking at a charity event within its newscast.

They're not allowed to use ScarJo's image with a caption saying "I trust CBS" on a billboard promoting CBS News.

Attack ads are part of political discourse, like the news. They're not being used to sell sneakers or cars.

Not to be snarky but companies like this are doing the lords work in that they're ensuring regulation comes fast and hard.
Good, advertising/movies isn't user generated art or satire. Depends on the country, but we have legal tests for that now.
> “What’s up guys? It’s Scarlett and I want you to come with me … ” she says, before the screen transitions into AI-generated photos that resemble her. A voice imitating the actor continues speaking to promote the app: “It’s not limited to avatars only. You can also create images with texts and even your AI videos. I think you shouldn’t miss it,” the fake narration says.

As AI gets better and better at this, how do we reasonably deal with the obvious ramifications? I'm sure some form of "this is me in this video / image" attestation has been given before, but that has its own downsides: namely, you have plausible deniability if you don't attest to that (e.g. if the video is of you being racist, etc.), plus how would the attestation process even work?

AIs imitating humans is genuinely scary, and I'm wondering what sort of technical / legislative safeguards are being put in place. I remember the case of a mobile app that allowed users to imitate Trump and Joe Biden; they were, of course, used to attempt to manipulate people into voting for a certain side.

Even if AI isn't perfect now, the only way is up. Soon, I'm betting it'll be very hard to discern what's AI-generated and what's real.

Really terrifying stuff.

> AIs imitating humans is genuinely scary

Eh, we got through photocopied signatures fine. The transition will be rough. But I wouldn’t get scared.

Now we have to get through the period of an AI generated version of you signing a document with a notary over zoom.
> I'm wondering what sort of technical / legislative safeguards are being put in place

None?

I don't think any new ones are needed. You don't need any special attestation. This company broke the law, ScarJo is suing them, and she'll win. The courts are well equipped to handle this with normal procedures, such as the simple fact that there is no contract signed with her.

As for political stuff, they can similarly be sued for libel/defamation/etc. And again, it's fairly trivial to figure out -- videos all come from somewhere, with a camera person who can testify, etc. If you have a video of Trump/Biden saying something that seems fake, and you suddenly "can't remember" where you got it, and it didn't exist on the internet until you posted it... it's really not that hard.

Photographs have been manipulated since 1846. It's pretty easy -- the more unusual or suspect something is, the more proof we demand (witnesses, corroboration, credibility, etc.).

The courts are well equipped to handle this with normal procedures

Be careful there. Your enlightened legal jurisdiction isn't all legal jurisdictions and some have some ideas we'd find pretty weird.

But yes, in the US, see: Lanham Act.

Your honor, we created an ai model of a blonde humanoid and named her scarlet and told it generate off of a movie star look. We then created a voice ai trained on people impersonating a well known actress.

I wonder how that will go in court.

Quite well for Scarlett, at least in the US (your jurisdiction may vary). It's established that if you get a look-alike actress, maybe a voice-over person who sounds like her, and make an ad implying it's actually her and she endorses the product, you lose[1]. Do you think AI makes this different?

[1] https://www.lexology.com/library/detail.aspx?g=c29600ca-8ef8...

"Conclusion: Celebrity endorsements in advertising have long been viewed as a better way to bring attention to a brand or product than an advertisement with unknown models or actors, which is why celebrities can command such a high price for their endorsements. Some celebrities even chose not to appear at all in any commercial advertising, which the right of publicity (and the right of privacy) empowers them to do. When companies are not successful in properly securing the legal right to use a celebrity’s image, identity, voice, or persona in an advertising campaign, they may be under the mistaken impression that it is nevertheless permissible to use a look-alike or sound-alike, in order to evoke the celebrity’s image in the minds of the public in connection with its own product, while not actually using the celebrity’s image. However, the right of publicity and the false endorsement branch of the Lanham Act give celebrities powerful tools to correct any such mistaken impression."

edit: pronoun clarity

Doesn’t look good for the defense. One difference I do see here is that there is an explicit disclaimer present on the ad if I am reading it correctly. That sort of disclaimer may work well for South Park but I don’t know about this.
Judges hate the "oopsie whoopsie did we do that? ;)" defense.
The solution is all real actors will have QR code formatted certificates on them at all times that you can scan to verify chain of trust.

Hey! Blockchain might actually be useful for something!

A qr code wouldn't do much...
People have been imitating celebrities for a very long time.
Good for her. Such obvious impersonation for gain strategies should not be encouraged.
Any link to the ad? Would love to see what it's about and how good it is
I'm sure there was no small part of this marketing strategy that was - at least as a worst-case scenario - hoping that people would share it in articles like this one.

Personally I say fuck that; let's not reward them.

AI kills the financial incentives to be creative. Its the death of uniquity.
How do you define "be creative"?
Whats the point in "being" Bob Dylan, when anyone can prompt AI to output the same? And if I were to innovate how to be the next Dylan, what would my motive be to continue to create once my innovation was absorbed into language models?
I hope her and ariana grande don't learn about the cyberpunk voice mods