Export
The link always opens the latest version of this page.
Ask an AI about it
Open in ChatGPTA new chat that reads this entry from the site Open in ClaudeThe same, in ClaudeShare the link instead of a file, and they will always see the latest version.
Files are a snapshot of today. The live page has anything newer.
AI-generated fake CVEs incident
Six bogus SQLite CVEs with CVSS 7.5–10.0 and 48 other fake vulnerabilities in libraw and ESP32-audioI2S were submitted via GitHub, passed through NVD/CISA enrichment, and exposed weaknesses in CVE validation.
A coordinated submission of 54 fabricated CVE entries through GitHub that successfully propagated through major vulnerability databases and scanning tools, exposing validation gaps in the CVE ecosystem.
In 2025, an unknown submitter via GitHub repository 1 created and submitted 54 bogus vulnerability records targeting multiple projects. Six fake CVEs were attributed to SQLite with severity ratings between CVSS 7.5–10.0, while 48 additional fraudulent entries targeted libraw and ESP32-audioI2S 1.
The fake CVEs passed through enrichment processes at both the NVD (National Vulnerability Database) and CISA (Cybersecurity and Infrastructure Security Agency) before being indexed in GitHub Security Advisories and incorporated into enterprise vulnerability scanners 1. None of the submitted vulnerabilities were reproducible 1.
The incident exposed significant weaknesses in CVE validation workflows. Despite multiple checkpoints in the vulnerability disclosure and enrichment pipeline, the fraudulent submissions were not filtered out before reaching downstream consumers including security professionals and enterprise security tools 1.
The primary victims were security professionals, vulnerability database operators, and organizations relying on enterprise scanners for vulnerability management 1. The pollution of authoritative vulnerability databases with non-existent CVEs created noise in security operations and potentially diverted resources from legitimate vulnerability remediation efforts.
This incident highlighted the need for stronger validation mechanisms in the CVE submission and enrichment process, particularly around reproducibility verification and source validation before entries are published to widely-trusted databases. The ease with which fabricated entries propagated through multiple authoritative sources demonstrated that the current ecosystem relies heavily on downstream trust rather than upstream verification.
Facts 4
| Attribute | Value | Confidence | Sources | Cited |
|---|---|---|---|---|
| Date | 2025 | low | 1 | 1 |
| Victim | security professionals, vulnerability databases, enterprise scanners | low | 1 | 1 |
| Impact | 54 bogus CVEs polluted NVD, CISA enrichment, GitHub Security Advisories, and enterprise scanners; none reproducible | low | 1 | 1 |
| Records affected | 54 | low | 1 | 1 |
Targets 1
Sources 1
- AI slop pollutes the CVE pipeline with fake vulns · The Register: Security