AI Science Tracker

AISLE AI-engine discovery of FFmpeg mbedTLS hostname verification bypass for IP hosts

FFmpeg · AI involvement attributed to AISLE AI-native discovery engine (specific model not identified)

Finding and attribution

In FFmpeg’s mbedTLS TLS backend, tls_open() skipped setting the expected hostname for IP-literal hosts. A network attacker able to intercept such a connection could present a trusted certificate for an unrelated name, because the client did not check the host against the certificate’s CN/SAN, and read or alter the stream. The issue is limited to the mbedTLS backend and IP-literal URLs; it is not a general FFmpeg TLS-backend bypass.

Product: FFmpeg. AI system: AISLE AI-native discovery engine (specific model not identified). Team / organization: AISLE / FFmpeg maintainers.

CVE identifiers: CVE-2026-107660.

AI and human contribution (source account): AISLE’s individual CVE page labels the issue “Discovered by AISLE,” and the CVE appears in AISLE’s index headed “CVEs our AI-native engine discovered.” The upstream FFmpeg commit credits Joshua Rogers as finder. Together, these support AISLE’s case-specific AI-engine discovery claim; no particular model, work split or independent priority is stated.

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 records a public disclosure date of 2026-09-06; the AISLE case page was published 2026-10-08. The cited sources do not report exploitation in the wild or establish zero-day status.

Remediation status: fix documented. The upstream fix sets the expected hostname for numeric hosts; the CNA identifies FFmpeg 8.1.3 and 9.0.2 as fixed branch cutoffs and links additional backports. Distribution-specific backports are not inferred.

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 6 Sep 2026. AI attribution dated 8 Oct 2026. Sources checked 17 Nov 2026; these are separate dates, not a live installation check.

Recorded CVSS: Medium · highest recorded score 6.3. Documented technical effects: data disclosure, data modification.

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.

  • CVE-2026-107660: CVSS 4.0 · 6.3 (Medium) · CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:L/VI:L/VA:N/SC:N/SI:N/SA:N. VulnCheck CNA · checked 17 Nov 2026.
  • CVE-2026-107660: CVSS 3.1 · 4.8 (Medium) · CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:L/A:N. VulnCheck CNA · checked 17 Nov 2026.

CVE-2026-107660: data disclosure, data modification.
Prerequisites / affected scope: Requires FFmpeg’s mbedTLS backend, an IP-literal host URL, and an on-path network attacker. When the hostname is not passed to mbedTLS, a trusted certificate for an unrelated name can be accepted, allowing the attacker to read or alter the HTTPS, RTMPS or TLS stream. Other FFmpeg TLS backends are outside this documented issue..
Exploitation evidence: The CNA describes an attack scenario; cited sources do not report exploitation in the wild.. Impact source · checked 17 Nov 2026.

Documented fix: The upstream fix sets the expected hostname for numeric hosts; the CNA identifies FFmpeg 8.1.3 and 9.0.2 as fixed branch cutoffs and links additional backports. Distribution-specific backports are not inferred.

Original evidence

Download this source record (JSON)

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