CVE-2026-65975

MEDIUMPre-NVD 6.56.5—
EchelonGraph scoreMEDIUM confidence

This medium-severity CVE scores 6.5 under a secondary CVSS source (NVD's own analysis pending). EPSS exploit probability: 0.2%, top 89% of all CVEs by exploit prediction. GitHub Security Advisory data not yet ingested — confidence will rise once GHSA publishes (typical lag: hours to days for open-source ecosystem CVEs; never for infrastructure-only CVEs).

Triggered by: NVD CVSS baseline
Sources: epss, secondary
Trending — 3 sources updated this week
6.5EG
EchelonGraph verdictMonitorLow exploitation likelihood right now — keep watching.
  • Lower severity and no public exploit yet
CISA-KEV: Not listedEPSS PROB: 0%CVSS: 6.5Exploit: None knownExposed: 0

No vendor fix yet — apply a workaround or compensating control (WAF / firewall / segmentation) and watch for a patch.

Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. In versions 1.88.0 up to but not including 1.107.1 and 2.0.0b1 up to but not including 2.5.0, the UI adapters (AG-UI via Agent.to_ag_ui()/AGUIAdapter, and Vercel AI via VercelAIAdapter) use sanitize_messages to strip unresolved ("dangling") client-submitted tool calls from untrusted message history before it reaches the agent, a defense-in-depth default that prevents the agent from executing tool calls the model never emitted. However, the strip anchored to a message index computed before sanitization ran, so when a trailing client message sanitized to empty and was dropped (for example a client system message under the default manage_system_prompt='server'), a preceding assistant response carrying an unresolved tool call became the new tail and was dispatched without inspection. As a result, a remote client could cause a registered, non-approval server tool to run with client-supplied arguments rather than arguments the model produced. The impact is bounded by what the affected tools do and is most significant for applications that gate tool execution in a model-request hook (before_model_request / after_model_request), since a forged call skips the model turn and bypasses that guardrail; approval-gated tools (requires_approval=True) are not auto-executed by this path. This issue has been fixed in versions 1.107.1 and 2.5.0.

CVSS v3
6.5
EG Score
6.5(medium)
EG Risk
44(Track)
EG Risk 44/100SSVC: Track

EG Risk is EchelonGraph's 0–100 priority score: it fuses intrinsic severity with real-world exploitation and automatability so you can rank equal-severity CVEs and fix the most dangerous first. Higher = act sooner. Distinct from the 0–10 EG Score (severity).

How it’s computed
Severity65% × 45%
Exploitation0% × 40%
Automatability100% × 15%
Action: Routine — remediate on your standard cadence.
EPSS PROB
0%
EPSS %ILE
11%
KEV
Not listed

Published

July 29, 2026

Last Modified

July 30, 2026

Advisory Details (1)

Auto-updated Jul 29, 2026
Patch available. Sources: github.
github Patch Available

UI adapters (AG-UI, Vercel AI): a dangling client-submitted tool call can execute when a trailing message is dropped during `sanitize_messages` · Advisory · pydantic/pydantic-ai · GitHub

https://github.com/pydantic/pydantic-ai/security/advisories/GHSA-jpr8-2v3g-wgf9

Weakness Classification(1)

MITRE Common Weakness Enumeration — the root-cause categories this CVE belongs to.

Data Freshness Timeline

(refreshed 7× in last 7d / 7× in last 30d)

Each row is a source pipeline that fetched or updated this CVE on that date, with what changed. For example, "NVD update" means NVD published or revised its analysis for this CVE; "MITRE cvelistV5" means we ingested or refreshed it from the CNA feed. Most recent first.

  1. 2026-07-30 19:09 UTCEG score recompute
  2. 2026-07-30 16:28 UTCEPSS rescore
  3. 2026-07-30 14:38 UTCEG score recompute
  4. 2026-07-30 14:08 UTCEG score recompute
  5. 2026-07-29 21:22 UTCEG score recompute
  6. 2026-07-29 20:25 UTCEG score recompute
  7. 2026-07-29 20:22 UTCMITRE cvelistV5first tracked

Related CVEs(same CWE)

Frequently asked(5)

What is CVE-2026-65975?
CVE-2026-65975 is a medium vulnerability published on July 29, 2026. Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. In versions 1.88.0 up to but not including 1.107.1 and 2.0.0b1 up to but not including 2.5.0, the UI adapters (AG-UI via Agent.toagui()/AGUIAdapter, and Vercel AI via VercelAIAdapter) use…
When was CVE-2026-65975 disclosed?
CVE-2026-65975 was first published in the National Vulnerability Database on July 29, 2026, with the most recent update on July 30, 2026. EchelonGraph re-ingests CVE updates from NVD on a 2-hour cycle, so this page reflects the latest published state.
Is CVE-2026-65975 actively exploited?
CVE-2026-65975 is not currently on CISA's Known Exploited Vulnerabilities catalog. FIRST EPSS estimates a 0% probability of exploitation in the next 30 days, which ranks it in the top 89.0% of all scored CVEs.
What is the CVSS score of CVE-2026-65975?
CVE-2026-65975 has a CVSS v3 base score of 6.5 (NVD).
How do I remediate CVE-2026-65975?
Patch to the fixed version published by the affected vendor. Where vendor advisories exist for CVE-2026-65975, EchelonGraph cross-links them in the Vendor Advisories panel below — those typically contain the canonical remediation steps, fixed version numbers, and any vendor-specific mitigations.

Dependency Blast Radius

Explore the affected products and dependency analysis for CVE-2026-65975

Explore →

Is Your Infrastructure Affected by CVE-2026-65975?

EchelonGraph automatically scans your cloud infrastructure and maps CVE exposure using blast radius analysis.

Start Free Scan →Browse All CVEs