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: 0.3% (23rd percentile of EPSS-scored CVEs). 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
6.5EG
EchelonGraph verdictMonitorLow exploitation likelihood right now — keep watching.
  • Lower severity and no public exploit yet
CISA-KEV: Not listedEPSS PROB: 0.3%CVSS: 6.5Exploit: None knownExposed services: Not assessed

A fix is available — apply it.

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
6.5MEDIUMmedium confidence
EG Risk
44
EG Risk 44/100CISA SSVC

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%
CISA SSVC: Track at low or medium mission impact; Attend at high (mission-essential systems).
Action: A fix is available. Apply it within your standard update timelines at low or medium mission impact and sooner than that at high.
EPSS PROB
0.3%
EPSS %ILE
23rd
KEV
Not listed

CISA SSVCTrack at low or medium mission impact; Attend at high (mission-essential systems).

A fix is available. Apply it within your standard update timelines at low or medium mission impact and sooner than that at high.

Exploitation none (CISA Vulnrichment) · Automatable yes (CISA Vulnrichment) · Technical impact partial (CISA Vulnrichment). Mission impact is CISA's Mission & Well-being decision point, and only you can judge it: high means the affected system is essential to your organisation's mission, or its compromise could cause irreversible harm to people. CISA's decision table

Published

July 29, 2026

Last Modified

August 4, 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 9× in last 7d / 39× 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.

Showing the most recent 100 of 105 total refreshes for this CVE.

  1. 2026-10-04 23:22 UTCEPSS rescore
  2. 2026-10-04 06:58 UTCEG score recompute
  3. 2026-10-03 14:26 UTCEPSS rescore
  4. 2026-10-02 17:57 UTCEG score recompute
  5. 2026-10-02 17:56 UTCEPSS rescore
  6. 2026-10-01 19:50 UTCEPSS rescore
  7. 2026-10-01 03:32 UTCEG score recompute
  8. 2026-09-30 15:03 UTCEPSS rescore
  9. 2026-09-29 07:32 UTCEG score recompute
  10. 2026-09-28 13:51 UTCEPSS rescore
  11. 2026-09-27 19:35 UTCEG score recompute
  12. 2026-09-26 15:59 UTCEPSS rescore
  13. 2026-09-25 05:30 UTCEG score recompute
  14. 2026-09-23 21:47 UTCEG score recompute
  15. 2026-09-23 17:54 UTCEPSS rescore
  16. 2026-09-22 15:45 UTCEG score recompute
  17. 2026-09-20 20:16 UTCEPSS rescore
  18. 2026-09-20 03:45 UTCEG score recompute
  19. 2026-09-19 04:45 UTCEG score recompute
  20. 2026-09-18 19:28 UTCEPSS rescore
  21. 2026-09-18 03:59 UTCEG score recompute
  22. 2026-09-17 05:15 UTCEG score recompute
  23. 2026-09-16 14:08 UTCEPSS rescore
  24. 2026-09-16 06:31 UTCEG score recompute
  25. 2026-09-16 05:15 UTCEPSS rescore
Show 75 more
  1. 2026-09-15 07:46 UTCEG score recompute
  2. 2026-09-14 02:42 UTCEG score recompute
  3. 2026-09-13 03:57 UTCEG score recompute
  4. 2026-09-12 05:12 UTCEG score recompute
  5. 2026-09-11 14:53 UTCEPSS rescore
  6. 2026-09-11 09:37 UTCEPSS rescore
  7. 2026-09-11 06:27 UTCEG score recompute
  8. 2026-09-09 07:44 UTCEG score recompute
  9. 2026-09-08 22:01 UTCEPSS rescore
  10. 2026-09-07 20:36 UTCEG score recompute
  11. 2026-09-07 16:01 UTCEPSS rescore
  12. 2026-09-06 13:47 UTCEPSS rescore
  13. 2026-09-06 10:18 UTCEG score recompute
  14. 2026-09-05 15:31 UTCEPSS rescore
  15. 2026-09-05 11:30 UTCEG score recompute
  16. 2026-09-04 11:53 UTCEG score recompute
  17. 2026-09-04 05:07 UTCEPSS rescore
  18. 2026-09-02 14:12 UTCEPSS rescore
  19. 2026-09-01 22:45 UTCEG score recompute
  20. 2026-09-01 13:54 UTCEPSS rescore
  21. 2026-09-01 04:40 UTCEPSS rescore
  22. 2026-08-30 21:51 UTCEG score recompute
  23. 2026-08-30 19:17 UTCEPSS rescore
  24. 2026-08-30 01:22 UTCEPSS rescore
  25. 2026-08-29 23:06 UTCEG score recompute
  26. 2026-08-28 21:42 UTCEPSS rescore
  27. 2026-08-27 09:41 UTCEG score recompute
  28. 2026-08-26 14:47 UTCEPSS rescore
  29. 2026-08-26 10:56 UTCEG score recompute
  30. 2026-08-25 13:49 UTCEPSS rescore
  31. 2026-08-25 12:11 UTCEG score recompute
  32. 2026-08-24 13:26 UTCEG score recompute
  33. 2026-08-22 15:56 UTCEG score recompute
  34. 2026-08-21 23:49 UTCEPSS rescore
  35. 2026-08-21 17:11 UTCEG score recompute
  36. 2026-08-20 22:56 UTCEPSS rescore
  37. 2026-08-19 19:38 UTCEG score recompute
  38. 2026-08-19 17:04 UTCEPSS rescore
  39. 2026-08-18 18:38 UTCEG score recompute
  40. 2026-08-18 13:48 UTCEPSS rescore
  41. 2026-08-17 19:53 UTCEG score recompute
  42. 2026-08-17 13:47 UTCEPSS rescore
  43. 2026-08-16 20:24 UTCEG score recompute
  44. 2026-08-16 14:56 UTCEPSS rescore
  45. 2026-08-15 21:40 UTCEG score recompute
  46. 2026-08-15 01:30 UTCEPSS rescore
  47. 2026-08-14 00:11 UTCEG score recompute
  48. 2026-08-13 22:00 UTCEPSS rescore
  49. 2026-08-13 01:26 UTCEG score recompute
  50. 2026-08-12 13:51 UTCEPSS rescore
  51. 2026-08-12 02:42 UTCEG score recompute
  52. 2026-08-11 03:58 UTCEG score recompute
  53. 2026-08-11 00:00 UTCEPSS rescore
  54. 2026-08-10 05:14 UTCEG score recompute
  55. 2026-08-09 13:47 UTCEPSS rescore
  56. 2026-08-09 06:29 UTCEG score recompute
  57. 2026-08-08 16:37 UTCEPSS rescore
  58. 2026-08-08 07:45 UTCEG score recompute
  59. 2026-08-07 09:00 UTCEG score recompute
  60. 2026-08-06 13:47 UTCEPSS rescore
  61. 2026-08-06 10:16 UTCEG score recompute
  62. 2026-08-05 19:17 UTCEPSS rescore
  63. 2026-08-05 19:17 UTCEPSS rescore
  64. 2026-08-05 11:31 UTCEG score recompute
  65. 2026-08-04 15:10 UTCEPSS rescore
  66. 2026-08-04 12:46 UTCEG score recompute
  67. 2026-08-04 10:39 UTCEPSS rescore
  68. 2026-08-03 14:12 UTCEG score recompute
  69. 2026-08-03 10:36 UTCEPSS rescore
  70. 2026-08-02 15:26 UTCEG score recompute
  71. 2026-08-02 02:27 UTCEPSS rescore
  72. 2026-08-01 16:42 UTCEG score recompute
  73. 2026-08-01 04:16 UTCEPSS rescore
  74. 2026-07-30 19:09 UTCEG score recompute
  75. 2026-07-30 16:28 UTCEPSS rescore

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 on July 29, 2026, with the most recent update on August 4, 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 0.3% probability of exploitation in the next 30 days (23rd percentile of EPSS-scored CVEs).
What is the CVSS score of CVE-2026-65975?
CVE-2026-65975 has a CVSS base score of 6.5 (a secondary CVSS source that NVD displays; NVD's own analysis pending).
How do I remediate CVE-2026-65975?
A fix for CVE-2026-65975 is available: update to the fixed version the vendor names in its advisory.

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.

Book a Demo →Browse All CVEs