AI Science Tracker

Valkey TLS connection use-after-free can permit remote code execution

Valkey · AI involvement attributed to Valkey LLM adversarial-audit pipeline (specific models not identified)

Finding and attribution

With TLS enabled, an authenticated client can issue CLIENT KILL against a connection while Valkey is iterating another TLS connection with buffered data, leaving a saved pointer to freed connection memory. The advisory says this can crash the server and that heap grooming may enable potential remote code execution. Non-TLS deployments are not affected.

Product: Valkey. AI system: Valkey LLM adversarial-audit pipeline (specific models not identified). Team / organization: Valkey project.

CVE identifiers: CVE-2026-56684.

AI and human contribution (source account): Valkey’s 2026-09-17 technical account describes the same TLS pending-data/CLIENT KILL use-after-free mechanism and states that this bug was found through adversarial testing audits run against the codebase using LLMs. The account describes a model proposing candidates and another model challenging, verifying and reproducing candidates before a person receives a reproducible crash. It does not identify the model or exact investigators for this CVE, nor establish which pipeline step produced this candidate.

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: Valkey’s 9.1.1 release publicly listed the CVE and fix on 2026-07-21; its GitHub advisory was published 2026-07-22. The AI-attribution account was published later, on 2026-09-17. No checked source reports exploitation in the wild or establishes zero-day status.

Remediation status: fix documented. Valkey 9.1.1, released 2026-07-21, lists the CVE and fix; the advisory lists patched versions 9.1.1, 9.0.5, 8.1.9, 8.0.10 and 7.2.14 for affected branches. This documents project release versions, not deployment or patch state in any installation.

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

Recorded CVSS: High · highest recorded score 7.5. Documented technical effects: code execution, denial of service.

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-2026-56684: code execution, denial of service.
Prerequisites / affected scope: TLS must be enabled. An authenticated client must be able to issue CLIENT KILL against a connection while another TLS connection has buffered data pending. Valkey notes heap grooming may be needed to reclaim and control the freed memory. Non-TLS deployments are unaffected..
Exploitation evidence: The advisory says the issue can crash the server and that controlled reclamation of the freed memory can potentially lead to remote code execution. It does not report exploitation in the wild.. Impact source · checked 17 Nov 2026.

Documented fix: Valkey 9.1.1, released 2026-07-21, lists the CVE and fix; the advisory lists patched versions 9.1.1, 9.0.5, 8.1.9, 8.0.10 and 7.2.14 for affected branches. This documents project release versions, not deployment or patch state in any installation.

Original evidence

Download this source record (JSON)

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