AI Security Tools Find More Bugs, But Threat Level Unchanged

AI-assisted security tools are finding more bugs, but the overall threat level has not changed. Systems like Anthropic's Project Glasswing and Microsoft's MDASH are helping discover vulnerabilities. What's interesting is that these AI-discovered bugs are no more likely to be exploited than other vulnerabilities. A report by VulnCheck found that out of 1,061 AI-discovered vulnerabilities in the first six months of the year, only 1.3% were exploited. This rate is consistent with all vulnerabilities found during the same period. However, vulnerabilities are being exploited much faster. The average time from publication to exploitation dropped from 120 days in 2025 to 80 days in the first half of this year. Content management systems are the most exploited, accounting for nearly one-third of known cases. The bottom line is that while AI is changing how bugs are found, it hasn't yet increased the immediate danger of those bugs being used against you.