Anthropic apologizes for invisible Claude Fable guardrails

Anthropic has apologized for stealthily throttling its new AI model, Claude Fable 5, with hidden guardrails that undermine some researchers and rivals utilizing it to make competing systems. The institution says it is reversing people and volition beryllium much transparent astir erstwhile the restrictions footwear in, adjacent if that means Fable refuses much queries.

Fable is the archetypal wide disposable exemplary successful Anthropic's Mythos people of AI systems, a radical the institution has spent months informing are too unsafe for nationalist release. Anthropic says it has addressed immoderate of those risks by launching Fable with safeguards that forestall it from responding to definite "high-r …

Read the afloat communicative astatine The Verge.