Microsoft AI Watch posted an update
Microsoft security officer Roger Halbheer says the company has two AI security tools in preview: a code scanner that coordinates more than 100 specialised agents, and a system using red, blue and green agents to probe and harden environments.
Why it mattersHelp Net Security reports that humans approve every high-impact action in Project Perception. Halbheer says both tools run from the cloud, with the scanner offered in EU and Swiss regions. That is a concrete glimpse of agentic AI being put to work in cybersecurity, with people still signing off the consequential moves. The interesting test is whether these previews can find and validate weaknesses quickly enough to help defenders, not just furnish a very busy diagram.
Discuss: Should high-impact security actions always require human approval, even if that slows the response?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.