JFrog exposes AI-generated fake SQLite CVEs

Automated CVE scoring and the rise of LLM slop threaten to overwhelm maintainers and undermine trust in security databases. Vulnerability databases blindly trusted plausible-sounding AI prose without verifying code existence or running proof-of-concept payloads. Maintainers are forced to waste valuable time and resources auditing phantom security flaws instead of patching actual code defects. CVE Numbering Authorities (CNAs) urgently need automated sandboxed PoC validation and strict reporter reputation systems to filter out AI-generated vulnerability spam.