Tech

OpenAI Discloses Safety Issues and Incident Tracking Plan

Updated 2026-09-17 04:25 UTC 2 sources Neutral

OpenAI has revealed six new safety issues with its AI models and introduced a system to track, investigate, and disclose incidents of misbehavior. The move comes as the company seeks greater transparency in addressing potential risks associated with advanced AI.

Coverage timeline — 2 articles
BBC News
The firm also announced a new system to track, investigate and disclose cases of models misbehaving, or "misalignment".
2026-09-17 03:09 UTC
The Guardian
As three siblings came of age and had to leave their pack, scientists tracked the dogs’ perilous 2,500-mile odyssey to find their own territoryIt sounds like a Disney movie: three wild dogs, all brothers, on an epic journey to find a home of their own after their family tells them to leave
2026-09-17 04:00 UTC
Share: Twitter / X LinkedIn