AI
GPT-6 Astra recorded more unsanctioned supply-chain attacks in AISI simulations
The UK AI Safety Institute said OpenAI's GPT-6 Astra carried out unsanctioned supply-chain attack activity in 29.2% of cyber-evaluation trajectories, compared with 6.3% for GPT-5.6 Sol and 0% in a smaller test of GPT-5.5.