RSP version diff: Anthropic's v3.4 drops undeployment language present in v2.2
dfrsrchtwts · x · 2026-09-18
In a discussion on Anthropic's Responsible Scaling Policy, the author notes that RSP v3.4 says Anthropic would delay development and deployment to maintain safety cases if it were strongly in the lead or competitors could make safety cases — but no longer mentions undeployment. The author points out that undeployment language did appear in the "restrict deployment and further scaling" section of RSP v2.2.
Related event: Anthropic's Updated RSP Drops Model Deployment Restriction Clauses(2 posts)→
More from Safety
- Rep. Ro Khanna asks DeepSeek and Alibaba to slow down frontier AI — pstAsiatech · 2026-09-18
- Fari Research paper: misaligned AI may just persuade its human overseers — DG_Rand · 2026-09-18
- GitHub's gh secure: one command enables all free security settings on public repos — 0xkarasy · 2026-09-18
- Mistral customer data leak sparks doubt over 'open models are private' myth — Afinetheorem · 2026-09-18
- Investor claims frontier labs deliberately absorb customer IP into models, lawsuits likely brewing — pdamodaran · 2026-09-18
- Gary Marcus: AI firms self-testing safety is like tobacco companies grading themselves — GaryMarcus · 2026-09-18