SignalTech/AI
Anthropic staff say AI could kill everyone. But its safety rule keeps getting weaker
Anthropic staff warn AI extinction risk while safety rules weaken.
Context
What happened: Anthropic staff have warned that AI could pose an existential risk, while the company's safety rule has been weakened. Anthropic now relies on filters that block weapons requests.
Why it matters: Until rivals slow down, the company has to decide whether a model is too dangerous to build, a choice shaped by competitive pressure.
What changed
Anthropic staff have warned that AI could pose an existential risk, while the company's safety rule has been weakened. Anthropic now relies on filters that block weapons requests.
Why it matters
Until rivals slow down, the company has to decide whether a model is too dangerous to build, a choice shaped by competitive pressure.