ThinkPatternGet the app
Story
TECHNOLOGY · NOV 14, 2026

AI Tools Accelerate Software Vulnerability Discovery

AI-assisted cybersecurity tools are rapidly identifying software vulnerabilities, though Google reports only 0.2% of these flaws have been exploited by hackers.

AI-assisted cybersecurity tools are significantly increasing the discovery of software vulnerabilities, a trend experts describe as progress rather than a precursor to a cyber apocalypse. Through Project Glasswing, Anthropic PBC identified over 129,000 verified vulnerabilities between April and July. Other AI tools have uncovered critical flaws in OpenSSL, FreeBSD, the PHP programming language, and the Linux kernel.

Microsoft Corp. demonstrated the efficiency of these tools by verifying Linux kernel flaws in approximately 22 minutes at a cost of $3.60 per flaw. Despite the surge in detections, a Google threat-intelligence report indicates that only 0.2% of these vulnerabilities have been exploited by hackers, suggesting defenders currently hold a temporary advantage.

A significant gap remains in remediation. While detection has accelerated, many organizations still take 30 to 60 days to fix critical flaws, and one study found that companies fully fixed only 26% of known-to-be-exploited vulnerabilities last year.

In response, recommendations for the U.S. government include expanding White House access to frontier AI capabilities for vetted defenders. Additionally, experts urge the adoption of the Cybersecurity and Infrastructure Security Agency's Secure by Design procurement rules to modernize federal software best practices.


Reported across 5 outlets
Actors
Anthropic PBCGoogle LLCMicrosoft Corp.Cybersecurity and Infrastructure Security Agency

Keep reading in the app

The full story and every source, free in the app.