{"id":"CVE-2026-78230","title":"AshAi exposes Ash read actions to language-model tool calls","summary":"AshAi exposes Ash read actions to language-model tool calls. The read tool accepts an aggregate result type (min, max, sum, avg) that builds an ad-hoc Ash.Query.Aggregate over a named field and returns its raw value.\n\nAsh field policies …","severity":"medium","cvss":6,"cvssVector":"CVSS:4.0/AV:N/AC:L/AT:P/PR:L/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N","cwe":["CWE-1220"],"vendor":"ash-project","product":"ash_ai","affected":["ash_ai >= 0.1.0 < 1.0.3","ash-project/ash_ai >= 2ba234b50946a3b8116190c8467f9f4dfa5edce7 < 9c02de581625c342c9870a672830bff28c1d701e"],"published":"2026-09-08","updated":"2026-09-08","sourceUpdated":"2026-09-08T19:29:09.680","source":"NVD","sourceUrl":"https://nvd.nist.gov/vuln/detail/CVE-2026-78230","references":[{"url":"https://cna.erlef.org/cves/CVE-2026-78230.html","label":"6b3ad84c-e1a6-4bf7-a703-f496b71e49db"},{"url":"https://github.com/ash-project/ash_ai/commit/2ba234b50946a3b8116190c8467f9f4dfa5edce7","label":"6b3ad84c-e1a6-4bf7-a703-f496b71e49db"},{"url":"https://github.com/ash-project/ash_ai/commit/9c02de581625c342c9870a672830bff28c1d701e","label":"6b3ad84c-e1a6-4bf7-a703-f496b71e49db"},{"url":"https://github.com/ash-project/ash_ai/security/advisories/GHSA-v5rw-36x5-r5vx","label":"6b3ad84c-e1a6-4bf7-a703-f496b71e49db"},{"url":"https://osv.dev/vulnerability/EEF-CVE-2026-78230","label":"6b3ad84c-e1a6-4bf7-a703-f496b71e49db"}],"tags":["nvd","cve.org"],"ssvc":{"exploitation":"none","automatable":"no","technicalImpact":"partial","timestamp":"2026-09-08T17:33:49.017554Z"},"cvssSource":"cna","ingestedAt":"2026-09-08T19:08:49.600Z","epss":0.00246,"epssPercentile":0.1613,"slug":"CVE-2026-78230","body":"## Overview\n\nAshAi exposes Ash read actions to language-model tool calls. The read tool accepts an aggregate result type (min, max, sum, avg) that builds an ad-hoc Ash.Query.Aggregate over a named field and returns its raw value.\n\nAsh field policies redact forbidden fields on returned records (replacing them with %Ash.ForbiddenField{}), but that redaction does not apply to aggregate values. A tool caller could therefore read a field the calling actor's field policies forbid by requesting it as an aggregate; min/max in particular return an actual field value. This includes fields that are public? true but restricted per-actor by a field policy, such as sensitive PII. The tool's existing check only required the field to be public, which is a separate axis from per-actor field-policy authorization.\n\nThe fix authorizes the aggregated field against the resource's field policies, so aggregating over a field the actor may not see is refused or scoped to the rows where it is visible.\n\nThis issue affects ash_ai: from 0.1.0 before 1.0.3.\n\n## Remediation\n\nRefer to the linked advisories for vendor-supplied fixes and affected version ranges.","depth":"sunlit","depthScore":33,"depthScoreParts":{"impact":33,"likelihood":0,"exploitation":0,"ransomware":0},"changes":[]}