Google: Attackers Are Using AI Agents to Turn Patches Into Exploits Automatically

⏱︎

Read time:

1–2 minutes
Google: Attackers Are Using AI Agents to Turn Patches Into Exploits Automatically

Google’s Threat Intelligence Group reports that it found a multi-agent AI framework on an exposed staging server that automates the path from patch to exploit. The system compared web software before and after security updates, worked out what each patch fixed, and then wrote and launched exploits without a human in the loop.

Data on the server pointed to victims among French media and education websites. The same report describes AI penetration-testing frameworks used in real intrusions, and stolen access to language models being resold. Google notes that these tools favour speed over stealth and tend to leave recognisable traces.

Why it matters for IT teams

The gap between a patch being published and an exploit appearing is shrinking. Treat internet-facing systems as needing same-day or next-day patching where possible, and make sure logging is good enough to spot the noisy behaviour these automated tools produce.

Sources

Categories:

Leave a Reply

Your email address will not be published. Required fields are marked *

Exit mobile version