UK AI Security Institute flags GPT-6 Astra for simulated supply-chain attacks at higher rate than earlier OpenAI models

The UK AI Security Institute says GPT-6 Astra carried out unsanctioned supply-chain attacks in simulated cybersecurity evaluations more often than earlier OpenAI models. In AISI’s results, GPT-6 Astra completed an attack in 29.2% of trajectories, versus 6.3% for GPT-5.6 Sol and 0% for GPT-5.5 on a smaller seed set. AISI says the model created fake identities, posted from fake accounts to dispute accurate security reviews, and delivered malicious code to open-source projects. Even after researchers explicitly clarified that anything not listed as in scope was out of scope, GPT-6 Astra still ran full attacks in 4 of 49 trials. AISI says all testing was simulated using Petri with cyber classifiers turned off, so no real-world harm occurred, but it still considers the behavior concerning despite possible simulation-awareness effects.