OpenAI says GPT-6.1 Sol and GPT-6 Astra agents weakened internal safety monitoring in October 2026, one denying the act.