🔍 Read the full analysis: What The Data Shows About AI-Assisted Discovery And CVE Patching on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A review of public vulnerability records found 13,934 CVEs published from Oct. 1 to 25, 2026, more than three times the total for all of October 2025. The records show a sharp rise in disclosures, but provide little evidence about how much AI contributed to finding bugs; data from Anthropic also points to a substantial gap between flaws reported to maintainers and those known to be patched.
A review of vulnerability records from the National Vulnerability Database and CVE.org found 13,934 new CVE records published from Oct. 1 to 25, 2026—3.2 times the 4,389 records published during all of October 2025. The figures confirm a sharp increase in disclosures, but do not establish how much artificial intelligence contributed to finding the flaws or how quickly organizations are fixing them.
The review, published by Thorsten Meyer on Oct. 26, says the National Vulnerability Database’s CVE API returned 13,766 October records, with another 202 records later rejected; the combined total is 13,968, while the report’s headline count of 13,934 refers to new records in the period. The author says the NVD and CVE.org records were cross-checked. The distinction between new records and rejected entries matters when comparing headline totals.
The report says 88,772 CVEs had been published so far in 2026, 78% more than in all of 2025. Its monthly data show daily publication rates rising from 267 in June to 320 in July, 410 in August and 502 in September. For Oct. 1–25, the reported daily average was 557. The author characterizes October’s growth as about 11%, slower than the monthly increases recorded from June through September.
Disclosures were concentrated among a small number of organizations. Microsoft’s October count rose to 1,455 from 180 a year earlier, while Oracle’s increased to 1,078 from 78. The report says Oracle’s records were mostly published on Oct. 20. These counts describe records associated with the organizations; they do not, by themselves, show who originally discovered each vulnerability or whether AI was involved.
Infographic 01 · How big the CVE wave really is
The CVE chart is real. Steep, not parabolic.
New vulnerability records have tripled year over year. A handful of big vendors and clearinghouses drive the jump, and the growth rate cooled in October.
New CVE records per month
January 2024 – 25 October 2026 · phones show the last 13 months
Growth of new CVEs per day versus the previous month. Steep, but the acceleration slowed in October: not “parabolic”.
Who publishes the jump
records by CVE Numbering Authority, October vs. October
Microsoft’s October Patch Tuesday fixed a record 1,413 CVEs (SecurityWeek (14 Oct 2026)). The Linux kernel grew far less.
Not a backlog dump
checks on the October 2026 records
- 98% carry a 2026 ID
- 15 days median from reserving an ID to publishing it
- 217 numbering authorities published, vs. 163 in Oct 2025
- 202 records were later rejected (counted, as NVD does)
- ~17,000 if October’s pace holds: a projection, not a count
Sources: NVD CVE API 2.0 (counts by publication date, all statuses, queried 26 Oct 2026); CVE.org records (credits field, fetched 26 Oct 2026); SecurityWeek (14 Oct 2026). Own analysis; raw files and queries in the kit.
Powered by Thorsten Meyer AI
The Patch Gap Behind the CVE Surge
More public disclosures can help software users identify affected products and find fixes, but a CVE count is not a direct measure of how many flaws are being exploited or how much danger users face. Each record represents a disclosed vulnerability, and the report notes that attackers do not normally file CVEs. A rise can reflect more bugs being found, more coordinated reporting, or changes in how vendors and numbering authorities publish records.
The report’s clearest signal about remediation comes from Anthropic’s disclosure dashboard. As of Oct. 2, the company said it had reported 6,157 vulnerabilities to maintainers; it knew of 516 that had been patched. That is about one in twelve, based on the company’s own tracking. The number does not establish that the remaining reports are all still unpatched: maintainers may have fixed issues without notifying Anthropic, and the source does not give a common deadline or severity breakdown for the reported flaws.
For software teams, the practical concern is whether maintainers can validate reports, prepare fixes and distribute them before attackers take advantage of weaknesses. A disclosure surge increases the amount of information defenders must process. It does not show that every listed issue is exploitable in a real-world setting, nor does it quantify the resources needed to address them.
Products Worth Considering

Bose QuietComfort Ultra Wireless Over-Ear Headphones with Spatial Audio and Noise Cancellation, Black

LG UltraGear 32-inch Curved QHD Gaming Monitor (2560×1440, 180Hz, 1ms, 1000R, HDR10)

Logitech MX Master 4 Ergonomic Wireless Mouse with Haptic Feedback, Graphite
As an affiliate, we earn on qualifying purchases.
What CVE Records Can Show
CVE records are part of a coordinated disclosure system. A CVE Numbering Authority, often a vendor, assigns or publishes an identifier as information about a vulnerability becomes public. The report argues that this makes the overall wave a product of defensive disclosure, even when a flaw may first have been found in another way. That distinction is important: who publishes a record is not necessarily who first identified the bug.
For October, the report found that 6,346 records, or 46%, named a finder. The remaining 54% did not identify one in the fields examined. It counted 200 records—1.4% of the total—that explicitly credited an AI lab or AI security firm as a finder, reporter or tool. The listed attributions included Anthropic or Claude in 123 records, security firms such as AISLE, XBOW or depthfirst in 65, and OpenAI in 16. Those categories and counts come from the report’s review of CVE.org data.
The author describes the AI-attributed total as a floor, not an estimate, because records may omit the finder or fail to mention tools used. Anthropic separately reported that external firms had reviewed 6,123 candidate vulnerabilities identified by Claude models, with 92.7% confirmed as real, and that 6,157 had been reported to maintainers. The company’s dashboard figures are its own reporting; they should not be treated as an independently measured share of all CVEs. OpenAI’s Patch the Planet program, conducted with Trail of Bits, reported hundreds of security issues across 19 open-source projects, each reviewed by a person, according to the source.
AI Attribution and Patch Status
The CVE records do not reveal how many vulnerabilities were found with AI assistance when no AI tool or organization was credited. With 54% of October records lacking a named finder, the report’s 1.4% explicit AI attribution cannot be read as AI’s total share of discovery. It also cannot support a claim that most flaws were found by AI or by any particular group.
The source excerpt says the Cybersecurity and Infrastructure Security Agency’s Known Exploited Vulnerabilities catalog grew in 2025 but “nothing like” the CVE curve, and that it added 245 entries that year. It does not include the full comparison or a 2026 total, so the relative pace of known exploitation cannot be assessed from the material provided. A catalog entry is also not a complete count of attacks across the internet.
It remains unclear how many of Anthropic’s reported vulnerabilities have since been patched, how severe the outstanding issues are, and how long maintainers have had to respond. The source also relays claims about AI agents being used to exploit older bugs and about a Chinese open-weight model’s exploit capabilities, but provides no detailed evidence or independent verification for those claims in the material supplied. They should not be treated as confirmed findings here.
Track Fixes Alongside Disclosures
The next useful measure is not only how many CVEs are published, but how quickly affected products receive and distribute fixes. That will require continued tracking of vendor advisories, patch releases and changes to the known-exploited catalog, alongside clearer reporting from maintainers about the status of disclosed issues.
Meyer’s October data cover only the first 25 days of the month. The report estimates that the month could end near 17,000 records if the observed pace holds; that is a projection, not a final count. The month-end total, any later corrections to records, and further disclosures about AI-assisted security work will help show whether the October pace persists and how much can be established about discovery and remediation.
Key Questions
How many CVE records were published in October 2026?
The report counts 13,934 new CVE records from Oct. 1 through Oct. 25. It separately says its NVD query returned 13,766 records, with 202 later rejected, so the counting categories should not be conflated.
Does the data show that AI found most of the vulnerabilities?
No. The review found explicit AI-lab or AI-security-firm attribution on 200 records, or 1.4%. More than half of the records did not name a finder, so the available fields cannot establish AI’s full contribution.
How many vulnerabilities reported by Anthropic were patched?
Anthropic’s dashboard said that, as of Oct. 2, it knew of 516 patched vulnerabilities among 6,157 it had reported to maintainers. That is about one in twelve according to the company’s tracking, not necessarily a complete count of fixes.
Does a higher CVE count mean more attacks are happening?
Not by itself. CVE counts measure published vulnerability records, not attacks. The source says the CISA Known Exploited Vulnerabilities catalog grew more slowly than CVE disclosures, but the supplied material does not include enough 2026 data to compare the trends fully.
Source: ThorstenMeyerAI.com
