AI-systemen vinden verborgen beveiligingsfouten in software

Leading cybersecurity teams demonstrated advanced AI bug-detection systems at DARPA's Artificial Intelligence Cyber Challenge in Las Vegas last August. The AI tools successfully identified most of the 54 million lines of intentionally flawed code, but notably discovered over a dozen additional bugs that DARPA had not planted. This development highlights the growing capability of artificial intelligence in identifying software vulnerabilities beyond human-designed test scenarios.