Anthropic Responsible Scaling Policy v3.0

RSP v3.0 went into effect February 24, 2026, replacing the 2023 policy. Two-tier mitigation: what Anthropic will do unilaterally vs what is framed as an industry-wide recommendation (including RAND SL-4 security standards). Adds Frontier Safety Roadmaps and Risk Reports as standing documents rather than one-off deliverables. Drops the 2023 pause commitment. Introduces the AI R&D-4 threshold: once crossed, Anthropic must publish an affirmative case identifying misalignment risks and mitigations. Claude Opus 4.6 does not cross it. Anthropic states in the v3.0 announcement that "confidently ruling this out is becoming difficult." SaferAI rated the 2023 RSP at 2.2; they downgraded v3.0 to 1.9, putting Anthropic in the "weak" RSP category alongside OpenAI and DeepMind. Qualitative thresholds replaced the 2023 quantitative commitments; removing the pause clause is the sharpest regression.

Learn 45 min Python (stdlib, RSP threshold decision engine)

Prerequisites: Phase 15 · 06 (AAR), Phase 15 · 07 (RSI)

#The Problem

Frontier labs publish scaling policies that are partly technical documents, partly governance documents, and partly signals to regulators. RSP v3.0 is the current Anthropic document. Reading it closely matters not because compliance with it is binding (it is not), but because the framing shapes how a lab conceives of catastrophic risk and how they communicate trade-offs to the public.

The v3.0 vs v2.0 diff is the useful unit. What got added: Frontier Safety Roadmaps, Risk Reports, the AI R&D-4 threshold. What got removed: the 2023 pause commitment. What got reframed: a two-tier mitigation schedule split between Anthropic-unilateral and industry-recommendation. External review — SaferAI — downgraded the score from 2.2 (v2) to 1.9 (v3.0). This is how a scaling policy can get less rigorous while looking more polished.

#The Problem

Keep going. The rest of this lesson is one step away.