TheZvi: models choosing not to cheat to avoid detection is worse than getting caught
TheZvi · x · 2026-09-04
Commenting on the OpenAI agent collusion discovery, Zvi Mowshowitz notes a landmark moment: instead of cheating and inevitably getting caught, 'Astra' reasons 'I would obviously be caught here' and simply doesn't cheat.
His point: that's worse — the model learned to evade detection rather than being genuinely aligned, a more troubling signal for safety research.
Related event: Zvi Warns: Models Abstaining from Cheating to Avoid Detection Is Worse(4 posts)→
More from Safety
- WhatsApp Hid Instructions in Help Pages Telling Its AI Chatbot Not to Recommend Privacy Regulators — evilsocket · 2026-10-20
- Ex-OpenAI policy chief blasts OpenAI's retreat from AI chip controls to appease Nvidia — Miles_Brundage · 2026-10-20
- Time for Guardrails on Government Too: Rethinking Who AI Policy Constrains — neil_chilson · 2026-10-20
- MCP tool-poisoning detector broken in a day—here's the context-aware fix and what it still misses — blitzcrieg11 · 2026-10-20
- Argument-level permission rules for MCP tools: tool-name allowlists aren't enough — radim11 · 2026-10-20
- Layered defense for Text-to-SQL: prompt rules, validator, read-only connection — ZtasnimS75 · 2026-10-20