Attack of the killer script kiddies

AI bug-finding tools are improving fast, and researchers say they can now uncover real software flaws and new vulnerabilities.

At DARPA’s AI Cyber Challenge and in later industry work, systems from groups like Anthropic, XBOW, Trail of Bits, and Theori demonstrated stronger offensive security capabilities.

The rise of tools such as Claude Mythos and Claude Opus 4.7 has heightened concern that attackers could automate exploit creation more easily.

The article warns that AI could lower the barrier to cyberattacks, letting nonexperts discover and weaponize vulnerabilities at greater speed and scale.