AI Science Tracker

FFmpeg: FFmpeg through 9.0.2 Double Free via cbs_sei_delete_message in cbs_sei.c

FFmpeg · AI involvement attributed to Claude and Ada Logics

Finding and attribution

FFmpeg through 9.0.2 contains a double free vulnerability in libavcodec/cbs_sei.c where cbs_sei_delete_message() leaves a stale extension_data reference that ff_cbs_sei_add_message() reuses. Attackers can supply a crafted H.264 stream with a display orientation SEI message followed by payload extension data, processed by h264_metadata with display_orientation=insert, to corrupt heap memory.

Product: FFmpeg. AI system: Claude and Ada Logics. Team / organization: VulnCheck (CNA); AI attribution as credited in the CVE record.

CVE identifiers: CVE-2026-125269.

AI and human contribution (source account): The CNA record credits 'Claude and Ada Logics; Joshua Rogers (AISLE Research); Michael Niedermayer'. It states the AI attribution as written there and does not describe the model's operation, autonomy or the human validation steps; none is inferred.

Human involvement is only described where the source account establishes it; unknown involvement does not mean none. Vendor confirmation of a flaw does not confirm every claim about AI contribution.

Discovery context: The CNA record establishes a coordinated disclosure with the credit above. It does not establish exploitation in the wild or a zero-day incident.

Remediation status: fix documented. CNA record links a fix commit or pull request; no fixed release version is stated in the record. Deployed patch status is not checked.

A source-documented fix does not establish installed patches or current safety. Source checking does not independently reproduce the finding or establish observed exploitation.

Disclosed 9 Nov 2026. AI attribution date not established by the source. Sources checked 17 Nov 2026; these are separate dates, not a live installation check.

Recorded CVSS: High · highest recorded score 7.3. Documented technical effects: Not available.

Severity, consequences and exploitation evidence

Case grouping uses the highest recorded CVSS score, with its version-specific severity band. It is descriptive, not a combined or editorial risk index. All recorded assessments remain below; scores from different versions and sources may disagree. CVSS specifications define version-specific scoring and severity bands.

Technical effects, prerequisites and exploitation evidence are not established here. Missing evidence does not establish absence of exploitation.

Documented fix: CNA record links a fix commit or pull request; no fixed release version is stated in the record. Deployed patch status is not checked.

Original evidence

Download this source record (JSON)

Open this case in the interactive tracker · Browse all static security cases