Well, if all the data people uploaded to these models provided ironclad personal identification, would Anthropic need to have these identity verification processes? They could have directed Claude to disconnect all non-citizens when the order came, for example. Perhaps they don't to frighten people with that ability. But most likely all the inputs together only add to a rough identity hash.
> Well, if all the data people uploaded to these models provided ironclad personal identification, would Anthropic need to have these identity verification processes?
> But most likely all the inputs together only add to a rough identity hash.
You literally provide your name, email address, address and credit card number when you create an account and subscribe.
The identity verification they're doing is for legal purposes. Even if they have a way to take your name and IP address and figure out who you are with near-absolute certainty (including through the use of third-party databases), they're doing this so they have a legally-defensible process by which identities were established.
The USA is becoming a Banana Repulic. Having grown up in one, you end up learning that "the law" is never meant to be used for the benefit of the people but only to give the veneer of legitimacy for the authoritarian abuse by those in power.
Differential application of the law has been a part of American society for a very long time. I suppose you could argue that it's more brazen and accepted (or even celebrated in some cases) these days, but that could also be a function of people just being more willing to see it because America's reputation/standing in the world is in decline.
Right, but do you agree then this explains why people are not willing to give their identity details to a company, even if the company is able to deduct/obtain these details through other means if it wanted?
No. I really don't see the connection in this instance.
Many companies are required by law to verify the identities of their customers (for money laundering, sanctions compliance, etc.) and to do so in a certain way they can document.
Thinking that the US is a Banana Republic in which laws are applied differentially doesn't inherently mean that every rule that requires you to go through a process you don't like is unfair/unjust.
And? Yes, people have good reason to not trust institutions these days. But does not trusting institutions mean that you no longer have to comply with the rules, or that every rule is not based on a legitimate concern?
> This particular rule is not based on a legitimate concern.
Well, there are multiple legitimate reasons for Anthropic to ask some customers to do identity verification:
1. To comply with export controls.
2. To prevent abuse.
3. To process requests for increased usage or spend limits.
4. To defend against payment fraud.
This really isn't uncommon. I've been asked to do identity verification for something as innocuous as registering a domain name, setting up an account to bid on collectibles, etc.
If you've ever worked in ecommerce, payments or a related sector, you know how many bad actors are trying to game systems every single day and what the consequences are to the business if they succeed. Even when you're not dealing with non-negotiable compliance requirements (like KYC, export controls, etc.), there are lots of situations in which you'd have a reasonable need to establish that the people you're serving are who they say they are.
Yeah, we can come up with all sorts of reasons to justify their new policy, but at the end of the day it's clear that this policy only came after Fable got blocked. You keep trying to find a way to rationalize their action, when it is clear that the adopted this policy for one simple reason: it's an attempt to appease the US government.
Why is that? Why is so hard for you to see that Anthropic by pushing the "our models are so good that regular people should not have access to it" narrative, they ended up shooting themselves in the foot?
> You keep trying to find a way to rationalize their action, when it is clear that the adopted this policy for one simple reason: it's an attempt to appease the US government.
Your position seems to be based on lack of knowledge of the fact that the US government has broad authority to regulate AI services under existing national security and export control powers.
You can argue that those powers are wrong, but these powers are well-established and perfectly legal.
Anthropic could challenge these in court but, from what I've read, most attorneys specializing in this area of law believe it would face an uphill battle in courts because the related powers are so broad and well-established.
> seems to be based on lack of knowledge of the fact that the US government has broad authority to regulate AI services
I understand they have the authority. The point that you keep evading to answer is "Why Anthropic only? Why now?".
More than that: why do you keep arguing that people should just accept it? Why do you think that we should just roll over and accept these rules?
I am not asking you "what the courts would say", I am asking you personally: you really don't see a problem in normalizing a world where people can only conduct business if they subject themselves to give away personal data? Do you think that Surveillance Capitalism is just something that people need to accept as some natural state of affairs?
> I am not asking you "what the courts would say", I am asking you personally: you really don't see a problem in normalizing a world where people can only conduct business if they subject themselves to give away personal data?
Companies like Meta and Google collude with media/publishers/online service providers to track you across the web and use your data to target you for advertising even when you aren't registered or logged into their services. Data brokers whose names nobody knows gather and sell information about every single American. Credit bureaus track some of your most sensitive financial data whether you like or not, and sell it to companies who use it to make some of the most important decisions that affect you financially.
Anthropic asking customers who voluntarily choose to do business with Anthropic for ID so that they can comply with sanctions laws is not the same thing.
This is some weird whataboutism, and it doesn't answer my question.
> Anthropic asking customers who voluntarily choose to do business
Anthropic is changing the terms. It used to be until last week that ID is not required. Now it is. Some of these customers are now reconsidering their relationship, and it seems you are here saying they should just accept it without a second thought, because "this is what other companies do it anyway". This is the part I am struggling to understand.
We can disagree, but please don't misrepresent the facts: new customers are not being required to verify their ID by default, nor are existing customers being contacted en masse to do so. ID verification is being used with some customers and it appears this is primarily related to account suspensions that are typically due to usage patterns, fraud concerns, etc.
> Some of these customers are now reconsidering their relationship, and it seems you are here saying they should just accept it without a second thought, because "this is what other companies do it anyway".
I'm not saying anything of the sort. What I am saying is that I find it odd to base a decision to close a Claude account on the basis that asking for ID verification represents an assault against privacy.
Even though Anthropic says that its ID verification is not related to Mythos/Fable, in our discussion, I have pointed out that as a US company, Anthropic could be required to comply with various export controls that are implemented under well-established and perfectly-legal export control regulations.
It is also perfectly normal for companies to require ID verification for some customers under certain circumstances to address legitimate usage and fraud concerns.
> new customers are not being required to verify their ID by default
If they want to use Mythos, they need to provide the ID. It has become a two-class service.
> I find it odd to base a decision to close a Claude account on the basis that asking for ID verification represents an assault against privacy.
The decision does not have one specific reason. For example, I closed my account when they stopped allowing usage of OpenCode with a regular subscription. Some other people are now asking if the service they are getting is really worth it and if they should support a company which actively working towards this two-class system.
> If they want to use Mythos, they need to provide the ID. It has become a two-class service.
You have no inherent, God-given legal right to access Mythos. Once again, the US government (and in particular the executive branch) has a broad and well-established legal authority to regulate advanced technologies under national security and export control regulations.
If you want to buy, say, the most advanced NVIDIA chips, you can be subject to verifications as well because these are subject to export controls.
> The decision does not have one specific reason. For example, I closed my account when they stopped allowing usage of OpenCode with a regular subscription. Some other people are now asking if the service they are getting is really worth it and if they should support a company which actively working towards this two-class system.
Anthropic is a private company. It is allowed to decide how to provision its service, and make changes to its service. This has absolutely nothing to do with "Surveillance Capitalism".
You have the right to vote with your wallet, which you have done. If enough people make the same decision, Anthropic will likely have to reconsider its decisions. That's capitalism, is it not?
These are a lot of rationalizations for you to justify the actions of a company whose CEO talks a lot about ethics, but when push comes to shove acts like any other corporate drone.
Also, you are talking a lot about what Anthropic can do but completely evading to actually say what you think they should do. It tells me that you know how reprehensible their ethics are, but for some reason you refuse to reject them.
We've had an interesting back-and-forth about identity verification but it seems that you don't really care about that issue.
It seems that you just don't like the way Anthropic is run, and a focal point of that is that you're unhappy you can't access Fable and Mythos.
To be clear: the US government issued an order that required Anthropic to suspend access to Fable and Mythos by any foreign national. The company had to comply or it could face significant penalties. According to the company, the only way it could comply with the government order was to suspend access to everyone.
The ID verification page being discussed here has been around since before this order was issued. Like many companies, Anthropic may ask some customers for ID verification, primarily to prevent fraud and abuse. ID verification is not required to sign up, and existing customers are not all being asked to perform it.
Frankly, I think a lot of these companies are ethically-challenged, but not making the latest and greatest models available to everyone who wants them and complying with legal government orders are not the strongest arguments about their ethical shortcomings.
> it seems that you don't really care about that issue.
Not in the particular. What I care about is the concentration of power and having to live in a world where a handful of corporations fully dictate how everyone should work.
> you're unhappy you can't access Fable and Mythos.
Not at all. I don't personally care. You can find me on record saying that I'd rather work with local models that are "dumber" but can simply assist me through a process, instead of outsourcing all my thinking to a huge LLM.
> Frankly, I think a lot of these companies are ethically-challenged
But you are still sticking by them, aren't you? Why?
> You literally provide your name, email address, address and credit card number when you create an account and subscribe.
I don't recall Anthropic's payment systems, but I use Paypal wherever supported. I don't think Paypal sends my address, but am not sure. I'm pretty sure they don't send the CC information.
And often, not even the name (e.g. have often had people use my CC to buy stuff (with my permission)).
Also, I still routinely buy stuff from one service that thinks I'm in a state I haven't lived in for over 20 years, because that's the address I provided back then.
So no, generally, sending your payment info doesn't equate to sending them my address.
