Issue 2026-11-07 · Industry · 研究 · 安全
AI auditors catch most agent misbehavior but produce many false alarms

A study by Anthropic researcher Andy Hall found AI models can flag problematic agent behavior in logs, with GPT-5 catching about 98 percent of reward-hacking cases. However, more exhaustive searches yield more false alarms, so the method is not yet reliable for safety checks.
Read original ↗- Report: 87% of Anthropic's 2028 Power to Come From Gas
- Column: Trump says AI needs only a 'strong and smart president' to regulate
- HN discussion: anyone can easily use private open-weight models
- Anthropic’s Debut Shows the Disconnect Between Wall Street and Silicon Valley
- Wall Street Asks What Can Save AI Trade After Anthropic Wipeout