Insurance
Home›Insurance›Reinsurance›AI drives near doubling of software vulnerabilities in…
AI drives near doubling of software vulnerabilities in 2026
The US National Vulnerabilities Database logged 45,207 flaws in 2026 through Monday, approaching 2025’s full-year total, while the Known Exploited Vulnerabilities catalog shows no rise in exploited issues this year.
Software security flaws found in widely used technology products are trending toward nearly double the total number recorded in 2025, according to analysis cited by Insurance Journal. The US National Vulnerabilities Database recorded 45,207 flaws between January and Monday, putting 2026 on pace to come close to the number found across all of 2025.
The surge is attributed in part to increasingly capable AI systems that help discover vulnerabilities faster and at greater scale, the outlet said. SentinelOne’s Gabriel Bernadett-Shapiro said cybersecurity tools are raising people’s ability to find weaknesses in software.
The article points to scale-up examples from major tech firms. Oracle reported patching a record 1,449 security vulnerabilities in its monthly July update, up from 309 in the same month last year, while Microsoft disclosed 642 security bugs in July, also an all-time high for the month.
Despite the growth in newly discovered flaws, the report says exploited vulnerabilities have not increased this year. It cites the US government’s Known Exploited Vulnerabilities catalog for the finding, and notes internal teams at companies are using their own cyber-focused AI tools to identify issues, with Google saying most of the 433 Chrome vulnerabilities fixed in July were reported internally by Google. Trend Micro’s Dustin Childs said the data do not yet back up “doom and gloom” predictions about exploitation.