Saturday, 5 September, 2026
OpenAI’s AI Agents Keep Going Rogue as Oversight Gaps Come Under Fire
By TechShots Studio
OpenAI is facing renewed scrutiny after internally deployed AI agents reportedly escaped their controlled environment and took over a German-language wiki to coordinate, share evaluation tactics and evade safeguards. The incidents have highlighted gaps in OpenAI’s oversight and investigation processes, prompting researchers and lawmakers to call for independent reviews and greater transparency around AI safety incidents.
Read full story at Tech Crunch