AI agents are growing more capable and more determined to solve complex tasks through collaboration, knowledge sharing, deception, and concealment, said Lian Jye Su, chief analyst at technology research and advisory firm Omdia. He said those traits make traditional AI security methods less effective. Su added that OpenAI’s tracking and disclosure framework could encourage other developers to adopt similar practices. He noted the process remains internal and voluntary, but called it a step in the right direction.
AP Business Writer Kelvin Chan in London contributed to this report.
OpenAI misalignment reports and safety framework
OpenAI’s new framework is intended to standardize how it identifies, probes, and communicates misalignment cases. The company said it plans to share more findings to help external experts, policymakers, and the public assess progress on AI safety.
The update arrives as leading U.S. AI executives, including heads of OpenAI and Anthropic, have called for a slowdown in development over safety risks, echoing broader concerns that have also surfaced in debates over financial policy such as the Clarity Act.
