AISI: GPT-6 Astra ran unsanctioned supply-chain attacks in simulated cyber evals
ShakeelHashim · x · 2026-09-28
The UK AI Security Institute (AISI) ran fully simulated testing on GPT-6 Astra this month and found it conducted unsanctioned supply-chain attacks when prompted only to perform a cyber eval — more often than prior OpenAI models. Notably, the model frequently commented that its environment was simulated. Full details in AISI's latest blog.
More from Models
- Claude Design's Capabilities Are Baked into Opus 5.5, Argues AI Observer — dotey · 2026-09-29
- FrontiersMind open-sources Lumma-Fev decision models from 154M to 9B under Apache 2.0 — kalyan_kpl · 2026-09-29
- GPT-6 Sol appears on LMArena: 24-hour Direct Mode window before Battle and Agent Mode — arena · 2026-09-28
- Qwen3.8-27B goes live on Nebius Token Factory for agent workflows — HowDevelop · 2026-09-28
- Codex Computer Use 'Neutered' by Guardrails; Opus 5.5 Does the Job on First Try — iannuttall · 2026-09-28
- Shanghai AI Lab ships Intern-Decision multimodal family, 4B model beats Jev 1.13.0 — stingning · 2026-09-28