This weekly update brings major news about keeping AI agents safe and secure. Companies are working hard to solve new problems that come with smart computer helpers.

Microsoft made big changes to its security system called Sentinel. The company wants to help security teams protect against cyber attacks using AI agents. These smart helpers can work faster than humans to find and stop bad actors. Microsoft's new system can understand connections between different parts of a company's computer network. This helps security teams see the big picture when attacks happen.

A group called Coalition for Secure AI talked about serious problems with AI agent identity. Right now, computer systems don't know how to treat AI agents. Are they like human workers or like computer programs? This confusion creates security risks. The group says AI agents need their own special type of digital identity. Without this, companies might have trouble controlling what their AI helpers can access.

The insider threat problem is getting worse with AI agents. These smart helpers might have access to sensitive company information. If something goes wrong, they could cause much more damage than a human worker. Security experts worry that AI agents could be tricked into doing harmful things. Companies need new ways to watch what their AI helpers are doing.

Several companies announced new tools this week. Dataminr launched AI agents that watch for threats in the real world. Entro created security tools specifically for AI agents. These companies see the growing need to keep AI helpers safe and under control.