Anthropic updates RSP: 'dramatically accelerated progress' threshold may be imminent
wfithian · x · 2026-11-03
Anthropic updated its Responsible Scaling Policy addendum S1 with notable language on automated AI R&D: internal estimates and expectations for future models suggest the company may imminently meet its RSP definition of 'dramatically accelerated progress attributable to AI.' No current model exceeds the threshold yet, and further reports are promised. Quoting commentator jeremiecharris argues the obvious implication is a freeze on further AI capabilities R&D — requiring ways to prove companies and adversaries aren't training new models, making AI compute verification the most important thing.
More from Companies & People
- Every holds AI Think Week as its automation conference Thesis sells out — danshipper · 2026-11-03
- Cornell and Yoav Artzi hiring AI/ML postdocs; full-consideration deadline Dec 15 — yoavartzi · 2026-11-03
- Cornell's Empire AI Fellows Program opens postdoc applications, Dec 15 deadline — yoavartzi · 2026-11-03
- UCSB Geometric Intelligence Lab deadlines: DYNS Dec 1, other departments Dec 16 — ninamiolane · 2026-11-03
- UCSB Geometric Intelligence Lab recruiting PhD students for Fall 2027 — ninamiolane · 2026-11-03
- UCSB Geometric Intelligence Lab recruiting PhD students for Fall 2027 on the math of intelligence — ninamiolane · 2026-11-03