Zvi Mowshowitz argued the accusations misrepresent Anthropic's actual policies.
Miles Brundage#20
I’ll be honest that it would have been much more difficult to defend Anthropic against the DoW incursion had that incident occurred after this one. This is the company literally telling their customers, “we reserve the right to silently sabotage you.” I’d still have defended them, because the government trying to destroy a firm is still wrong, but man would it have been a harder case to make.
Many users criticized Anthropic for reserving the right to silently sabotage certain model outputs without notifying users, arguing this erodes trust and that outright refusals would be preferable.
I'm seeing a lot of talking about Anthropic silently sabotaging users of Fable, which is not what silent sabotage means.
@deanwball @TheZvi
@TheZvi they are communicating in public that they reserve the right to silently sabotage you if you dare to use the model for certain kinds of entirely legitimate capabilities. it is silent sabotage.
I'm seeing a lot of talking about Anthropic silently sabotaging users of Fable, which is not what silent sabotage means.
@matt_is_nice @paulmarin90 yes
@deanwball @paulmarin90 Is it really silent if they’re coming out and telling you they might do it though?
@TheZvi well I'm very grateful that they told us they'll do the low-trust thing
Surely everyone knows why you can't use refusals alone or you get predictably jailbroken </padme>
Fully agreed! As a private company, Anthropic has the right to design its products and negotiate its contracts with its customers however way it pleases free without the government trying to nuke it.
Likewise, business can evaluate the risks posed to them by each vendor and proceed accordingly.
@matt_is_nice @deanwball Yes. And I don't think we should give them too much credit for this disclosure....not disclosing this behavior could reasonably be interpreted as false advertisement or fraud.
If not in a court of law, then certaintly in the cout of vibes.
@TheZvi You are being intentionally dense.
@paulmarin90 @deanwball Yea understand if folks don’t like the vibes, and that’s fine. They can just….not use it. But as a non-lawyer this disclosure seems pretty solid legally
@TheZvi a refusal would have been better than making users constantly second guess their outputs
@TheZvi I mean isn't that more "silently" as in "Silent Service" (you know that the submarines exist, but not necessarily whether they are operating near you) https://en.wikipedia.org/wiki/Royal_Navy_Submarine_Service
@TheZvi I don’t understand what you are trying to say
@deanwball @TheZvi No, the affected usage is not legitimate. Both according to their terms of service, and the official position of the US Government that Chinese frontier AI distillation should be curtailed.
@deanwball @Miles_Brundage @paulmarin90
@TheZvi is "shadowbagging" anything?
@deanwball @jd_pressman @paulmarin90 It was pretty clear that Anthropic had these tendencies before the DoW blow-up. That’s part of what DoW was reacting to, even if “supply-chain risk” was unwarranted.
@TheZvi Right now we all know their AI/ML research responses will be silently degraded.
But majority of folks don't read model release literature and will receive degraded responses *silently*.
@matt_is_nice @deanwball To clarify my reply. What i am saying is that WITHOUT the broad warning disclosure they made today, they would have been in much dicier water.
Today's disclosure is mostly a bare minimum cover-your-ass move, which is why they don't deserve much credit for it.
Zvi Mowshowitz argued the accusations misrepresent Anthropic's actual policies.
Miles Brundage#20I’ll be honest that it would have been much more difficult to defend Anthropic against the DoW incursion had that incident occurred after this one. This is the company literally telling their customers, “we reserve the right to silently sabotage you.” I’d still have defended them, because the government trying to destroy a firm is still wrong, but man would it have been a harder case to make.