← Tuesday, August 4, 2026
AI Frontier · Jul 27
ModerateMicrosoft's new AI hunts for security holes and fixes them
Microsoft built its first AI model made just for cybersecurity. It reads large amounts of code to find weak spots that hackers could use. It works with a team of AI agents: some act like attackers, some spot threats, and some fix them. The system, called Project Perception, opened to more testers on August 3.
Why it matters to GCs
Builders and infrastructure firms are real targets for hackers, and attacks on utilities are rising. AI that finds and fixes weak spots fast could help protect your project data and job site systems. But the same tools can help attackers, so keep your guard up.
TechCrunch·Via POLR AI Hub