00:00U.S. officials are warning of a new phase in cyber risk driven by artificial intelligence.
00:04Treasury Secretary Scott Besson and Fed Chair Jerome Powell recently convened Wall Street
00:08leaders over concerns about a powerful new AI model from Anthropic. It's called Mythos Preview
00:13and it may represent a turning point. Anthropic says the system can autonomously find and exploit
00:18software vulnerabilities, including so-called zero-day flaws, which are unknown to developers
00:23and highly valuable to hackers. In some cases, these bugs have gone undetected for decades.
00:29The model can also chain together multiple weaknesses to gain full control of systems,
00:34compressing complex vulnerability research into minutes, often without human supervision.
00:39That capability is why access is tightly restricted. Under a program called Project Glasswing,
00:44companies like Microsoft, Alphabet, and NVIDIA are using the tool defensively, racing to identify
00:49and fix vulnerabilities before attackers do. But risks remain. Less than 1% of the flaws identified
00:54so far have been fully patched. And as AI accelerates both defense and attack, the window to respond
01:00is shrinking. Anthropic believes AI will ultimately favor defenders. But for now, the transition may be
01:05volatile and increasingly difficult to control.
Comments