Cybersecurity defenders have officially entered uncharted territory as automated bug-hunting tools push annual vulnerability disclosures past 100,000 for the first time.
VulnCheck counted 100,222 CVEs published in 2026 as of Nov. 13, more than double the previous year’s total, according to The Register. For security teams, the challenge is identifying which disclosures affect their systems and require urgent action.
"There's no denying the fact that vulnerability hunting with AI is driving a majority of what we're seeing," Josh Bressers, VulnCheck’s director of coordinated vulnerability disclosure, told The Register.
The AI acceleration
Major CVE Numbering Authorities (CNAs), including Microsoft, Oracle, GitHub, and VulnCheck, are publishing flaws at record volume. Microsoft released 1,094 vulnerabilities in its November update after logging 1,474 in October.
However, disclosures and known exploitation are not growing at the same pace. VulnCheck’s first-half analysis found:
Measure | Finding |
| CVE publication volume | Increased 45% over the preceding six months |
| New known exploited vulnerability additions | Increased 10% over the preceding six months |
| Median time from CVE publication to exploitation | Fell from 120 days in 2025 to 80 days in H1 2026 within the analyzed dataset |
More published CVEs do not establish how many attacks occurred. Exploitation evidence can also emerge months or years after disclosure, making recent exploitation totals incomplete.
From January through June 2026, CrowdStrike found that 88% of its observed exploitation of vulnerabilities with public proof-of-concept code occurred within 48 hours of the code’s release.
VulnCheck’s Copeland XWEB EVO disclosure documents exploitation attempts against two of 14 disclosed vulnerabilities. For one flaw, its sensors recorded attempts about one hour and 40 minutes after the CVE record became public on Nov. 12.
Beyond the counting game
Crossing six figures reinforces why vulnerability management needs more than severity scores. Exploitation evidence, asset exposure, and business impact also matter.
In Microsoft’s November release, an Outlook spoofing vulnerability with a CVSS score of 6.5 was reported under active exploitation, illustrating why a medium score can still warrant urgent action.
AI-assisted discovery also creates work for maintainers who must validate reports and prepare fixes. Anthropic’s Project Glasswing illustrates the expanding use of AI for defensive research, but candidate findings should not be equated with confirmed vulnerabilities or published CVEs.
Separately, curl ended its bug bounty program on Jan. 31 amid concerns about low-quality submissions. From Oct. 26, the project also moved low- and medium-severity vulnerability reports to public GitHub issues, reserving private handling for high- and critical-severity reports.
"We don't have a security problem anymore," Bressers said, per The Register. "We have a data problem now."
What security teams should prioritize
The growing disclosure count makes a repeatable triage process more important. Teams need to establish which vulnerabilities affect deployed systems, which have credible exploitation evidence, and what an attacker could reach.
Start by connecting vulnerability findings to an asset inventory. A CVE should lead to an affected product and version, a system owner, and a remediation decision.
- Confirm exposure: Identify vulnerable internet-facing applications, VPN gateways, email servers, and management interfaces. Include connected operational equipment that may sit outside routine IT scanning.
- Use exploitation intelligence: Check CISA’s Known Exploited Vulnerabilities catalog, VulnCheck KEV, and vendor advisories alongside CVSS scores. Absence from a catalog does not establish that a flaw is safe.
- Assess business impact: Prioritize systems that provide privileged access, hold sensitive information, or support essential operations. Document why an affected system needs immediate action or can follow a scheduled patch cycle.
- Mitigate delays: When patching must wait, apply vendor workarounds and restrict vulnerable interfaces. Remove direct internet exposure from unsupported OT and IoT equipment, and coordinate replacement with operational owners.
- Verify the fix: Confirm that updates reached affected systems and resolved the vulnerable condition. Give unresolved exceptions an owner, review date, and documented mitigation.
Track how long exposed, exploitable systems remain unresolved, alongside patch completion. Closing tickets is useful only when the team can verify that the underlying exposure has been removed.





