Anthropic could have EASILY made an in-house solution with < 100 employees. But no, they have to make the same mistakes that behemoths like IBM and Oracle make. Y’all haven’t even IPO’d yet and we’re already entering the “sleepwalk to your grave” phase of the corporate cycle.
Flock (YC S17)
Flock is a Y Combinator company.
Isn't the case for any US company? I'm working in an european SaaS business, and we consider that any data that goes through Azure or Google Workspace can be accessed by the US government on a whim, even if their datacenter is on our soil.
You might also consider shipping software-as-software, which doesn't need cloud resources, and the data can only be searched one customer at a time.
They are literally being forced to do this by the government. The counter-ask if for them to self immolate in order to take a principled stand, as access to all their models is banned.
I'll pick my poison thanks and choose Chinese surveillance.
Assuming they won’t trade or sell it back to the US or whoever?
Seems to be working for every other AI company.
It feels like people here are not able to separate their personal dislike of the effective altruist types from a real debate on how big the stakes of AI are.
They don't care if anyone is using a fake ID, they care that they can send the hot potato back to the American government. That gives them a legitimate "not my fault if people can make fake ids. In fact it's your fault"
This help page has been there for a long time. Long before Fable was announced.
Their ID process isn’t new.
They might use it in the future to gate access to Fable if they can conclude that it’s sufficient to comply with the regulation, but the fact that they had this process in place already and they’re not using it for the Fable situation suggests there’s more to it.
How would they know? Does the US government expose their database to private businesses? I always wondered how these things work.
I normally run away from these ID checks. It’s just a matter of time until lone of these databases with everyone passports and videos get hacked. And vibe coding only makes it more likely.
I don’t care how good Claude is, it’s not good enough for that kind of risk.
I'm rooting for China here, feels strange but well
There is much stronger IDV technology than this, but it isn't as consistent across the customer base. You would have a significantly more difficult time defeating something like Chexsystems with fake IDs. If the AI models are truly so scary as to require ITAR-style restrictions, then I wonder if having an adverse financial background might be a reasonable canary for preventing access. These same kinds of questions come up for obtaining security clearances.
I noticed this is normalized too, there’s a systematic discrimination just based on the fact where you were born, as if anyone controls that, and yes, you will be added on some list, get denied or further scrutiny based on that fact only.
