Funny how no-one talks about AI neutrality like we used to discuss net neutrality. We literally now enter a space where not only you will have to prove your identity with a gov issued ID, but they will silently block you if they deem you try to use it in a way that they don't like.
It is literally similar to a situation where your ISP would investigate all sites you visit and limit your bandwidth if they don't like the the ones you enter...
> It is literally similar to a situation where your ISP would investigate all sites you visit and limit your bandwidth if they don't like the the ones you enter...
Same? Multitude of magnitues worse. The amount of data and type that is given here is from different level. ISPs have mostly seen it in encrypted format.
THe reason why its not being talked about like net neutrality is because there aren't large companies with a vested interested interest in changing the status quo
Net neutrality was about Google/netflix/etc not wanting to pay for transit to verizon/AT&T/etc
Same with the copyright reforms, new, richer internet companies (at the time) wanted to avoid paying feesto copyright owners.
The morality of these campaigns are out of scope, the point is, ID checks align with the new money.
> Net neutrality was about Google/netflix/etc not wanting to pay for transit to verizon/AT&T/etc
That is the dishonest spin put forward by the "last-mile" ISP providers.
The ISPs were already being paid by their customers in order to access the internet, but they wanted to leverage their natural monopolies in order to be paid twice, by whoever their customers were connecting to. Famously, the ISPs went as far as artificially throttling traffic in order to extort payments[1].
Framing this as a battle instigated by google and netfix is also the opposite of the truth. It was the ISP monopolies who brought this fight; not the internet companies. The battle was actually instigated by one of the last-mile ISPs, Verizon, suing the FCC.
> not wanting to pay for transit to verizon/AT&T/etc
If anyone was going to pay for "transit to" the last mile ISPs, they would actually be paying backbone companies like L3, because those are the ones that actually move traffic across the internet.
[1] the throttling typically took the form of throttling certain backbone connections, so that no one would be able to prove they targeted Netflix specifically
OpenAI also has this kind of check. What is especially bad is that if you fail the verification process, they won’t let you retry – you are permanently locked out from the top models. They aren’t clear about this upfront during the process, so make sure the lighting is good when you scan your ID!
Thankfully, OpenAI's check is not required if you use their models through OpenRouter!
This didn't use to be the case (OpenRouter's OpenAI access used to be bring-your-own-key), but they've reached some sort of deal with them a couple months ago, and now you can access all the GPT-5 series models on OR with no verification at all.
Why would they care about _you_ when they have just a bit short of a billion users and they are up for a huge IPO? Of course they won't even bother implementing a retry.
Worth noting that China implemented mandatory real-name verification for generative AI services back in 2023. The practical effect wasn't just about preventing misuse -- it created a two-tier system where verified users get full capabilities while others get heavily restricted outputs. What's interesting is how quickly the market adapted: local open-source models partly flourished because they sidestep these requirements. Western providers are now walking a similar path, but without the digital identity infrastructure China already had in place.
Every time I consider renting a service from Anthropic, they drop such bomb. Full capability with pre-agreed price per token and no ID verification. That's what I demand.
I love openrouter for this, I just put in $20 and i’m able to chat with almost every model out right now or plug their api into any ide that supports openai api requests. I use llm’s off and on and it’s nice not worrying if i’m getting my use out of a subscription. just note that claude subscriptions can be a lot more cost effective vs paying per token if you’re a power user.
I really don't like that headlines have been surfacing about the US government putting pressure on Anthropic, and now a short time later they are requiring ID's (albeit for certain use cases, but that's a slippery slope).
I may very well stop using Claude due to this.
Also, who is providing the verification service? We don't want another Discord situation.
EDIT: Just saw it's Persona. Definitely dropping Claude now.
I just hope there's huge protest. I hope people just cancel the subscription. I fear people will just accept the terms. The result will be a kill switch for the US government and a clean distinction between national and foreign users so spying will become legal. Surely Anthropic hasn't allowed the NSA connect so far, - openai clearly did (see their board members).
to play devil's advocate, having a credit card does not tell Anthropic anything about your country of citizenship, which the US is pushing them to gate keep access on.
For certain capabilities one can use uncensored models which can be found on huggingface. It's perfect for asking on how to create atomic bombs, meth labs, assassination plans, brute force hack scripts and more. You only need one or two h200 cards.
I mean, this feels exactly what had to happen after their announcement with Fable being restricted by the US government since the requirement is that they need to know you're a US citizen. You can argue this is Anthropic's fault due to their Mythos/Fable fear mongering, but at the end of the day this is a requirement by the US government to use this (and likely future models).
I expect to see this repeatedly with new powerful models from all providers.
Best I can do is root for local models (already was), but I'll keep my Anthropic subscription for their "lesser" models without an ID (for now).
One dimension of this which isn't discussed enough is this opens the road to inference providers silently discriminating against different users who will remain oblivious to what's going on. i.e. if you "fail" ID verification it's actually good that they tell you as opposed to serving you a malicious model instead.
This might seem unrelated but on one of my free accounts. I tried to make Claude do some historical fact checking on the inter-war period of the USSR. The point isn't if it is true or not, but it felt like it would help quite a lot to see what the sources Claude finds says about the both sides of the picture and I was curious at some point.
Funnily enough, a day after this my account got banned under the pretext that I was a child using Claude and that I would need to verify my account. The age verifier said that it doesn't store my photos or anything. It gets cheeky though and indirectly it says it doesn't store what I upload but sends it to third parties that do store and sell it.
Its like saying I won't steal your money, but I will give it to the thief right over there for free.
Now the flagging might be entirely coincidental, but I just exported my chats and just never went on with the intention to re verify my account (since it is a free one basically and there is no incentive for me to do so).
Weirdly enough, I started to see my past chat history that I exported to check and see if there is any correlation between how I talked and if there might have been some instances in which the system might attribute said message as what a young person would say. Though from the looks of it, it didn't give any of that sort of vibe.
179 comments
[ 4.0 ms ] story [ 105 ms ] threadIt is literally similar to a situation where your ISP would investigate all sites you visit and limit your bandwidth if they don't like the the ones you enter...
Same? Multitude of magnitues worse. The amount of data and type that is given here is from different level. ISPs have mostly seen it in encrypted format.
Net neutrality was about Google/netflix/etc not wanting to pay for transit to verizon/AT&T/etc
Same with the copyright reforms, new, richer internet companies (at the time) wanted to avoid paying feesto copyright owners.
The morality of these campaigns are out of scope, the point is, ID checks align with the new money.
That is the dishonest spin put forward by the "last-mile" ISP providers.
The ISPs were already being paid by their customers in order to access the internet, but they wanted to leverage their natural monopolies in order to be paid twice, by whoever their customers were connecting to. Famously, the ISPs went as far as artificially throttling traffic in order to extort payments[1].
Framing this as a battle instigated by google and netfix is also the opposite of the truth. It was the ISP monopolies who brought this fight; not the internet companies. The battle was actually instigated by one of the last-mile ISPs, Verizon, suing the FCC.
> not wanting to pay for transit to verizon/AT&T/etc
If anyone was going to pay for "transit to" the last mile ISPs, they would actually be paying backbone companies like L3, because those are the ones that actually move traffic across the internet.
[1] the throttling typically took the form of throttling certain backbone connections, so that no one would be able to prove they targeted Netflix specifically
https://help.openai.com/en/articles/10910291-api-organizatio...
what did you do to trigger a verification process?
This didn't use to be the case (OpenRouter's OpenAI access used to be bring-your-own-key), but they've reached some sort of deal with them a couple months ago, and now you can access all the GPT-5 series models on OR with no verification at all.
> At this time, retries are not supported. You can continue using OpenAI’s platform with your existing access.
That's ridiculous, especially as their list of reasons that verifications can fail include "There was a technical issue during submission".
Why would they care about _you_ when they have just a bit short of a billion users and they are up for a huge IPO? Of course they won't even bother implementing a retry.
Indeed, that's what motivated me to get an OpenRouter account!
I may very well stop using Claude due to this.
Also, who is providing the verification service? We don't want another Discord situation.
EDIT: Just saw it's Persona. Definitely dropping Claude now.
They’ve had this process in place for a long time.
This page is not new or recent. Someone just found it and submitted it again. It’s been discussed on HN months ago if you want to go look.
As mentioned in that thread, Persona as the provider is a bit surprising and problematic.
Discord dropped them after user backlash.
Not it's not. Evil companies like to do business together.
Never been a better time to use local models.
https://claude.ai/settings/billing?action=cancel-refund
I expect to see this repeatedly with new powerful models from all providers.
Best I can do is root for local models (already was), but I'll keep my Anthropic subscription for their "lesser" models without an ID (for now).
Funnily enough, a day after this my account got banned under the pretext that I was a child using Claude and that I would need to verify my account. The age verifier said that it doesn't store my photos or anything. It gets cheeky though and indirectly it says it doesn't store what I upload but sends it to third parties that do store and sell it. Its like saying I won't steal your money, but I will give it to the thief right over there for free. Now the flagging might be entirely coincidental, but I just exported my chats and just never went on with the intention to re verify my account (since it is a free one basically and there is no incentive for me to do so). Weirdly enough, I started to see my past chat history that I exported to check and see if there is any correlation between how I talked and if there might have been some instances in which the system might attribute said message as what a young person would say. Though from the looks of it, it didn't give any of that sort of vibe.