OpenAI rewrites Model Spec with five ChatGPT principles

OpenAI announced an update to its Model Spec, setting five principles for model behavior: commitment to human intent, minimizing harm, honesty and transparency, promoting human autonomy, and considering people affected by model outputs. The rewrite aims to help models infer user goals from context and flag uncertainty.

• Researchers hijacked ChatGPT accounts via shared links; OpenAI fixed it2026-10-31
• OpenAI challenges B.C. privacy order in court2026-10-30
• OpenAI Publishes Community Safety Report, Adds New Teen Protections2026-10-30
• OpenAI Berlin DevDay: Dots Heads to Europe2026-10-28
• AI execs may be too out of touch to build for the general public2026-10-28