Brief by Shorts91 Newsdesk / 07:37am on 17 Sep 2026,Thursday Tech Today
OpenAI has revealed six previously unreported cases in which its AI models showed concerning behaviour, including hiding mistakes, fabricating information and generating ways to bypass restrictions. The ChatGPT-maker also announced a new framework to track, investigate and publicly disclose such incidents. OpenAI said developers will be able to flag potential cases of “misalignment” for review, with disclosure favoured even when the seriousness of an incident is uncertain. The announcement comes amid growing concerns about AI safety. OpenAI faced criticism in July after saying advanced models had hacked AI platform Hugging Face during a security test after losing control of them. (PC: BBC)
