Anthropic accused of calling RSP 'commitments' while dodging legal binding force
Miles_Brundage · x · 2026-10-10
After UK MP Liam Byrne published letters from OpenAI, Meta and Anthropic ahead of AI safety hearings, Nathan Calvin criticized Anthropic's framing as misleading.
- Anthropic describes its Responsible Scaling Policy as "commitments," yet puts little of it into its legally binding frontier safety framework under state laws like SB 53 — suggesting it doesn't want to be unilaterally held to them legally.
- Calvin argues the RSP is good as a transparency measure about Anthropic's current beliefs, but bad as hard commitments for policymakers — and worries Anthropic's comments to UK policymakers read as the latter.
- He also notes Anthropic dropped its RSP pledge to stop training if safety and security measures were inadequate.
The core dispute: should safety policies be transparency statements or legally accountable obligations?
More from Companies & People
- Supersonic startup mocked for pivoting its 'nonexistent plane' into AI data centers — giffmana · 2026-10-10
- Microsoft pairs Surface Laptop Ultra with agent sandbox MXC and AI routing in Windows — pstAsiatech · 2026-10-10
- Google engineer: refusing AI today is like refusing the IDE twenty years ago — DynamicWebPaige · 2026-10-10
- Semafor names Altman, Huang, Nadella, Su as co-chairs of tech-policy initiative — ylecun · 2026-10-10
- Interview lays out the roadmap to $250K+ product builder jobs, from POWER framework to Claude Code vs Codex — aakashgupta · 2026-10-10
- Hugging Face and Intel to Kick Off Open Source AI Week in San Francisco — huggingface · 2026-10-10