Anthropic: Models can automatically improve safety benchmarks without degrading capabilities
AnthropicAI · x · 2026-08-29
Anthropic had Claude "hill-climb" safety benchmarks for common misalignments like deception or sycophancy, with the constraint of preserving general capabilities. The best methods generalized to held-out benchmarks and reliably improved safety scores.
Related event: Anthropic's Autonomous Alignment Researcher Outperforms Human Experts(22 posts)→
More from Safety
- FTC subpoenas Anthropic and OpenAI over safety incidents and consumer protection concerns — asusarla · 2026-10-16
- Anthropic red team's first-ever model "abliteration" cut GLM-5.3 refusals from 90% to 3% for ~$4,400 — cephaloform · 2026-10-16
- Anthropic grants METR employee-like access as independent safety evaluators go public — AnthropicAI · 2026-10-16
- APEX-Harm benchmark: script-based injection attacks easily compromise agents in shared workspaces — KAUST · 2026-10-16
- Residual-stream forecasts predict agent violations before they happen, USC study shows — USC · 2026-10-16
- Blogger doubles down: Anthropic's Scion image model accused of ripping off Mononoke — sunshowers · 2026-10-16