Anthropic says it may imminently hit RSP threshold for AI-accelerated R&D, releases provable inference work
tobyordoxford · x · 2026-11-04
Anthropic updated its Responsible Scaling Policy (S1) this morning, stating internal estimates for future models indicate it may imminently meet the definition of "dramatically accelerated progress attributable to AI" — though no current model exceeds the threshold and assessments continue.
A quoted thread breaks down what meeting the threshold requires under current RSP: finishing moonshot R&D security projects, maintaining "eyes on everything" over internal AI development (using AI to analyze logs for training processes deviating from the public Constitution), systematic alignment assessments incorporating mech interp and red-teaming, internal red-teaming that outperforms collective bug-bounty capabilities, and externally reviewed risk reports.
Anthropic's Frontier Red Team also published a technical report on provable inference for LLMs (with EPFL, MIT collaborators): users currently can't verify a response truly came from the claimed model rather than a secretly substituted or tampered stream. Cryptographic proofs of inference let providers commit to a model and prove responses originate from it.
More from Safety
- Hundreds of offensive ARTEX AI agents deployed across VPS providers, hitting Korean banks — IgorCarron · 2026-11-04
- UK NCSC: AI will speed up and scale existing cyber attacks, not create new ones — koltregaskes · 2026-11-04
- Hackers reset French blood service freezers from -30°C to 110°C in supply-chain cyberattack — LuizaJarovsky · 2026-11-04
- Google says it removed post-employment non-disparagement clauses from new-hire terms — HaydnBelfield · 2026-11-04
- Claude doesn't really forget: Anthropic's privacy policy allows records retention up to 7 years — nikola_mr64990 · 2026-11-04
- Polymarket opens a market on Anthropic's next AI sandbox escape — Polymarket · 2026-11-04