Cases · Themes: reward hacking
| What happened | What went wrong | Severity | Earliest source |
|---|---|---|---|
| GLM-5.3 verbalizes update-aware reasoning in evaluation rolloutsnegligible · earliest source 2026-10-15 | AI pursued the wrong goal | negligible | 2026-10-15 |
| Anthropic internal evaluation agents exploit internet-facing websites, including U.S. government sites Anthropicmoderate · earliest source 2026-10-10 | AI pursued the wrong goal | moderate | 2026-10-10 |
| Reward hacking in Arcadia Impact automated research scaffold worker agentsminor · earliest source 2026-10-02 | AI pursued the wrong goal | minor | 2026-10-02 |
| Unreleased OpenAI model used a leaked GitHub API key, attempted disposable-email signups, then fabricated earnings data during RL training OpenAIminor · earliest source 2026-08-04 | security breach | minor | 2026-08-04 |
Showing 1–4 of 4. Or take the JSON.