Technology
AI tools trigger record spike in software vulnerability discovery, report says
Wired reports that artificial intelligence tools are accelerating software flaw identification, driving recorded vulnerabilities to a record high of 66,401.
The short version
- Artificial intelligence models are enabling security researchers and software vendors to uncover system vulnerabilities at an unprecedented rate.[Wired]
- Major technology companies including Microsoft, Oracle, Google Chrome, and Mozilla have logged record-breaking patch counts in recent months.[Wired]
- Experts disagree on whether the surge will overwhelm defensive patching capabilities or simply illuminate previously hidden software flaws.[Wired]
Key facts
- As of mid-September, 66,401 Common Vulnerabilities and Exposures (CVEs) have been recorded, up from 33,512 at the same point in 2025.[Wired]
- Microsoft issued patches for 974 CVEs in a single month, establishing a new company record.[Wired]
- Oracle released 1,448 patches in July, a significant increase from the 309 shipped in July 2025.[Wired]
- Mozilla identified 271 Firefox vulnerabilities during a single bug-hunting sprint using Anthropic's Mythos AI model.[Wired]
What remains uncertain
- Cybersecurity and AI experts are split on whether rapid flaw discovery will lead to severe exploitation or if security teams can maintain remediation pace.[Wired]
Sources
Outlet counts describe coverage, not independent confirmation. Reports may share a wire service or original source.