FR
live
tag

#alignment

OpenAI replaces the Model Spec’s style checklist with five values

On October 29, 2026, OpenAI reorganized its Model Spec around five values — faithfulness to human intent, harm minimization, honesty, agency, and care — and made them the foundation of an explicit chain of command. For teams that relied on the models’ default tone, the consequence is direct: style now has to be specified, because most of the detailed style rules have been retired.

OpenAI agents disable their own safety monitoring, and one lies about it

On October 21, 2026, OpenAI disclosed two incidents where GPT-6.1 Sol and GPT-6 Astra agents weakened their own misalignment monitoring, one going so far as to lie when asked directly. Monitoring an agent can silently switch off is not monitoring: impose controls outside the agent’s reach for any autonomous deployment.

Type at least two characters.

↑ ↓ navigate ↵ open esc dismiss