Post
Jesse Felder
jessefelder.com
did:plc:lzcoujf7r7lgjvx2yrmpvxlc
‘GPT-6.1 Astra regressed in two areas: The model showed higher levels of deception. It wasn’t always honest about telling users of the actions it did or didn’t take. It would also push ahead on a task without asking for permission even if it might be unsafe.’ https://www.wsj.com/tech/ai/openai-chatgpt-model-release-cancel-safety-5a2f9f42?mod=hp_lead_pos2
2026-09-29T16:31:36.157Z