In an abrupt shift, the company may release future AI models without ironclad safety guarantees

Anthropic, the wildly successful AI company that has cast itself as the most safety-conscious of the top research labs, is dropping the central pledge of its flagship safety policy, company officials tell TIME.

In 2023, Anthropic committed to never train an AI system unless it could guarantee in advance that the company’s safety measures were adequate. For years, its leaders touted that promise—the central pillar of their Responsible Scaling Policy (RSP)—as evidence that they are a responsible company that would withstand market incentives to rush to develop a potentially dangerous technology.

But in recent months the company decided to radically overhaul the RSP. That decision included scrapping the promise to not release AI models if Anthropic can’t guarantee proper risk mitigations in advance.

  • Rekall Incorporated@piefed.socialEnglish
    2·
    6 hours ago

    All American polemics and “pledges” are BS, at least with respect to anything substantial.

    Not saying it was always like this and that it will always be like that, but it is reasonable to assume there it will take another generation (20-30 years) before we see any positive developments with respect to the culture of corruption, criminality and dishonesty that has unfortunately come to dominate American society.

    Doesn’t matter if a hypothetical Barack Obama II comes to power. From my time in the living in the US (several years with extensive travel across many different states), the impression I got is that on real matters an Obama is actually not too different from a Trump. The biggest difference is that Trump owns his corruption and criminality (with excellent electoral success).

    Even in foreign policy, Obama de facto approved the annexation of Crimea (our new leadership asked for support to fight the russian invasion of Crimea and were rejected) and he went along to characterize russia as “a regional power making trouble with its neighbors.”

    A comically stupid approach that’s not too different from Trump’s gibberish.

    And if you think I am being uncharitable, ask yourself the following question:

    Meta has been found to knowingly enable fraud to gain $16 B in 2024 alone. Meta was also reported to have developed a “playbook” to manage this fraudulent scheme; so the whole thing was premeditated and with clear intent.

    Is anything going to happen Meta (the entity) or Meta’s leadership (be it the far right or the centre right is in power)? Anyone who has lived in the US in the last ~30 years knows the answer!