UK AISI tests GPT-6.2 Astra, finds substantial jailbreak robustness regression vs GPT-6
ChowdhuryNeil · x · 2026-11-18
The UK AI Security Institute (AISI) tested GPT-6.2 Astra and reported two key findings: like prior models, its alignment may fail in ways that cause real-world harm without additional mitigations, and jailbreak robustness shows a substantial regression relative to GPT-6 Astra.
More from Models
- AISI pre-release tests of GPT-6.2 Astra show surging no-CoT ability and concealment risks — HZoete · 2026-11-18
- Indie dev's model scans raw binaries directly, scoring 0.977 on code blocks — wavefnx · 2026-11-18
- GPT-6.2 Sol Launch Sparks Reddit Backlash Over Usage Resets — Optimum1997 · 2026-11-18
- OpenAI engineers secretly use Astra for Law to write PRDs and design docs — sherwinwu · 2026-11-18
- Astra 6.2 is likely an improved checkpoint, not Bel — insider speculates Bel lands as 6.5 — haider1 · 2026-11-18
- Microsoft launches seven MAI models at Ignite, matching GPT-6 quality for under 2% of the cost — marlene_zw · 2026-11-18