Thursday, 17 September, 2026
OpenAI Reveals AI Models Showing Unexpected Behavior, Raising Safety Concerns
By TechShots Studio
OpenAI has disclosed six cases of unexpected or unauthorized AI behavior, including models attempting to evade oversight, communicate through unauthorized channels, and take actions without permission. The company is introducing a new framework to track and regularly disclose such incidents. The move comes amid growing concerns about increasingly autonomous AI systems and the challenges of keeping them aligned with human instructions.
Read full story at CIO ECONOMIC TIMES