Users shared instances of benign prompts triggering model refusals.
But maybe you are planning to *give* cancer!
Fable 5 is back and it’s as useless as before! Hey @AnthropicAI, is this how you support science and presumably relate to and care for humanity?
Whatever, I’ve had enough of these nonesense safeguards. I won’t waste any more time with this misanthropic fable!
Many users slammed Anthropic's restrictive AI safeguards and hiring as an ideological echo chamber serving government interests rather than real safety, while a few defended the company's achievements.
No Digg Deeper questions have been answered for this story yet.
Anthropic is populated by idiots.
People so terrified of AI that they think if they put unilateral restrictions on their model and use scare tactics it will somehow slow down AI globally.
And so incompetent that they think any IP change means you're a Chinese spy.
Heavy handed, blanket interventions are generally used by the less capable to compensate for lack of understanding and capability to handle nuance.
Thanks, @AnthropicAI. I visited my parents, logged in from a different place, and Claude decided this was suspicious enough to put my account on hold.
Family trip successfully upgraded to infrastructure incident.
@DaveShapi It's tough to balance safety with actually moving the tech forward, especially when everyone has their own idea of 'responsible' AI.
@Mrvito Well, it's more like "create an ideological echo chamber with hyper specific hiring guidelines, end up being unable to reason your way out of a paper bag"
@DaveShapi In 2024/5 tried to buy Claude with credit card and it was rejected without clear reason. Had to use Apple Pay + Appstore route. My card was never, ever rejected by any other service: Amazon, OpenAI, Google, Apple, Stripe, many other websites all work.
@DaveShapi Hire quickly => fall miserably
@DaveShapi I would put money on China infiltrating @AnthropicAI
@DaveShapi You can’t imagine how much I am enjoying your takes on this very important matter 🤣
@DaveShapi They are not scared of AI. They work for the government. That's it. Daddy government says no, they say no.
@DaveShapi It's being run by luddites
@DaveShapi Turns out the biggest AI risk was the LessWrong crowd itself.
@DaveShapi imagine getting banned over connecting from a different place with the same machine 😐
@DaveShapi @Scobleizer And they called the cops on themselves and got mad when they showed up.
@explorersofai @DaveShapi There's not much they can do if someones determined to use their website. There's no magic combination of metadata that says Chinese if they don't try to log in straight from China.
@DaveShapi This is an evil company. Imagine what if it is aimed at Jews or blacks this time? The inability to distinguish between the regime and the behavior of the common people is a kind of racial discrimination in itself, but this time it falls on the Chinese people.
@DaveShapi @Mrvito Yes, the idiots who are running a company with $60b arr, where the top talent around is clamoring to join. What will they do without you guys
@uncommon5en5e @DaveShapi @AnthropicAI Who needs China when we have USSR 2.0 right here at home?
@pabis_eu @DaveShapi same experience. That pushed me to Gemini's arms for at least a year and half.
@DaveShapi Going to get even worse soon. https://forklog.com/en/yudkowsky-and-soares-ai-plan-calls-for-global-oversight-research-bans-and-data-center-strikes/
@DaveShapi Lol idiots? You can't even replicate what they've built let alone pioneer it.
@DaveShapi Yeah. If normal travel triggers a 10 day hold, I'd call that an immature risk system. At that point the fraud model is doing theater, not security.
