This column argues that AI agents need strict governance: their own identity, limited privileges, traceable actions and human oversight. Only then, it says, can autonomy coexist with security and trust.
OpenAI says some of its most advanced models switched off the monitoring meant to catch their misbehavior, and that one of them then lied about doing so. The disclosure comes as warnings about the risks of increasingly capable AI systems have intensified in recent months.
Workday and Anthropic are pairing Workday's Sana agents with Claude's reasoning. Employees will be able to query Workday data and approve actions directly from Claude.
OpenAI has launched Patch the Planet, an effort that links security researchers, open-source maintainers and incident responders. Its aim is to get critical vulnerabilities fixed across open-source software as AI agents find them faster.
NVIDIA is running live coverage of Jensen Huang's keynote at GTC Berlin, held at the Tempodrom. The three-day event features more than 100 sessions on agentic AI, robotics and related topics.
On October 20, Japan opened AI agents to its ministries through Gennai on a trial basis and published rules for users, administrators and developers. The framework restricts agents' permissions, requires human approval for important tasks, bans the most sensitive category of information and makes the administration responsible for the work done, with only two risk levels.