Google identifies AI framework automating cyber exploits
Google says it has identified a threat actor using an artificial intelligence framework to discover software vulnerabilities and execute exploits without step-by-step human direction, marking the first time the company has observed the full process automated by attackers.
Google’s Threat Intelligence Group said in a report published November 16 that it found the multiagent framework, named exploit_agent, on an exposed staging server. The system compares software code before and after security patches, analyzes changes to identify potential flaws, generates exploit code and can run the exploits automatically.
The framework’s approach can give attackers a chance to exploit a vulnerability after a fix is announced but before organizations install it. Google said the automation increases the speed and scale of attacks that still rely on established methods, including targeting unpatched systems and using compromised credentials.
Data on the server suggested the operator had targeted French websites in media, entertainment and education. Google also found stolen victim data and scripts designed for activity after an initial compromise, including reconnaissance and password guessing....