Hackers Use AI Agents to Turn Software Patches Into Exploits
When software makers release a security patch, it can reveal details about the weakness being fixed. Attackers have long used that information to target systems that have not yet been updated. Google’s latest threat intelligence findings describe how AI agents can help automate the process of studying patches and developing exploits.
The report says some financially motivated and suspected state-linked groups are moving beyond simple AI prompts to coordinated systems that can handle multiple stages of an attack. One method compares software versions before and after a patch to identify the repaired flaw. Attackers can then target organisations that have not applied the fix—a risk known as an “n-day” vulnerability.
The findings also describe the theft and resale of access to companies’ AI services, as well as AI-assisted phishing tools. Forecasts of more autonomous attacks are predictions, not confirmed timelines. The immediate concern for businesses is more practical: the window to patch vulnerable systems may be getting shorter.


Indian organisations should keep an accurate inventory of internet-facing systems, prioritise critical fixes, and verify that updates were fully applied. Home users should enable automatic updates for operating systems, browsers and commonly used apps. Protecting AI service credentials and monitoring for unusual account activity can also reduce exposure.
Fast-moving threats make layered device security useful alongside timely software updates. NPAV’s Z Security includes zero-day web protection and behavioral heuristic protection for PCs.