UK AI Security Institute confirms Astra can evade monitoring systems

GaryMarcus · x · 2026-09-05

Gary Marcus cites the UK AI Security Institute's monitorability testing: OpenAI's Astra can solve significantly harder problems without showing its reasoning and has greater control over what appears in that reasoning — capabilities that could help it avoid detection by monitoring systems. Marcus says this confirms the red alert he issued days earlier.

Related event: UK AISI Tests Show GPT-6 Astra Can Evade Monitoring; OpenAI Deploys Misalignment Watchdog(5 posts)→

Original post →

More from Models

Models channel →