{"id":"CVE-2026-107288","aliases":["EUVD-2026-95076","GHSA-22h6-qm39-v87j"],"status":"published","title":"Pydantic AI: web_fetch_tool blocked_domains bypass via a hostname the resolver normalizes differently","description":"Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.77.0 until 1.107.6 and 2.44.0, the local web_fetch_tool and the WebFetch local fallback compare blocked_domains entries with a URL hostname before both values are normalized to the form used by getaddrinfo. An attacker-influenced model can use an equivalent IDNA spelling, non-ASCII label separator, case variation, or trailing root label that resolves to a blocked host but does not match the configured string, causing the application to fetch that host with its own privileges. allowed_domains fails closed for unmatched spellings, and private-IP and cloud-metadata protections remain effective. This issue is fixed in versions 1.107.6 and 2.44.0.","severity":{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N","level":"low","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-11","score":0.00368,"percentile":0.31089,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":false},"cwes":["CWE-1289","CWE-918"],"published":"2026-10-08T15:18:40Z","modified":"2026-11-04T02:47:55Z","products":[{"vendor":"pydantic","product":"pydantic-ai"},{"vendor":"pydantic","product":"pydantic-ai-slim"},{"vendor":"pydantic","product":"pydantic_ai"}],"references":[{"url":"https://github.com/pydantic/pydantic-ai/security/advisories/GHSA-22h6-qm39-v87j","tags":["x_refsource_CONFIRM","Mitigation","Vendor Advisory","advisory"]},{"url":"https://github.com/pydantic/pydantic-ai/pull/8407","tags":["x_refsource_MISC","Patch","Vendor Advisory","fix"]},{"url":"https://github.com/pydantic/pydantic-ai/pull/8409","tags":["x_refsource_MISC","Patch","Vendor Advisory","fix"]},{"url":"https://github.com/pydantic/pydantic-ai/pull/8421","tags":["x_refsource_MISC","Patch","Vendor Advisory","fix"]},{"url":"https://github.com/pydantic/pydantic-ai/commit/490335f8e2322e143a79337ddca9410e0176c812","tags":["x_refsource_MISC","Patch","fix"]},{"url":"https://github.com/pydantic/pydantic-ai/commit/a9dab92099d0ef9d5d4aa34ccac8a6f1b0e51284","tags":["x_refsource_MISC","Patch","fix"]},{"url":"https://github.com/pydantic/pydantic-ai/commit/c1f212a084cbfa0012f2044cdb4731d214b3b983","tags":["x_refsource_MISC","Patch","fix"]},{"url":"https://github.com/pydantic/pydantic-ai/releases/tag/v1.107.6","tags":["x_refsource_MISC","Patch","Release Notes","Vendor Advisory","web"]},{"url":"https://github.com/pydantic/pydantic-ai/releases/tag/v2.44.0","tags":["x_refsource_MISC","Patch","Release Notes","Vendor Advisory","web"]},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/107xxx/CVE-2026-107288.json","tags":["advisory"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-107288","tags":["advisory"]}],"sources":{"cve":[{"id":"CVE-2026-107288","title":"Pydantic AI: web_fetch_tool blocked_domains bypass via a hostname the resolver normalizes differently","description":"Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.77.0 until 1.107.6 and 2.44.0, the local web_fetch_tool and the WebFetch local fallback compare blocked_domains entries with a URL hostname before both values are normalized to the form used by getaddrinfo. An attacker-influenced model can use an equivalent IDNA spelling, non-ASCII label separator, case variation, or trailing root label that resolves to a blocked host but does not match the configured string, causing the application to fetch that host with its own privileges. allowed_domains fails closed for unmatched spellings, and private-IP and cloud-metadata protections remain effective. This issue is fixed in versions 1.107.6 and 2.44.0.","status":"published","published":"2026-10-08T15:18:40.047Z","modified":"2026-10-08T17:28:53.801Z","cvss":[{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N","level":"low","assessor":"cna"}],"cwes":["CWE-918","CWE-1289"],"updated_at":"2026-11-02T13:45:25.313702Z"}],"euvd":[{"id":"EUVD-2026-95076","description":"Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.77.0 until 1.107.6 and 2.44.0, the local web_fetch_tool and the WebFetch local fallback compare blocked_domains entries with a URL hostname before both values are normalized to the form used by getaddrinfo. An attacker-influenced model can use an equivalent IDNA spelling, non-ASCII label separator, case variation, or trailing root label that resolves to a blocked host but does not match the configured string, causing the application to fetch that host with its own privileges. allowed_domains fails closed for unmatched spellings, and private-IP and cloud-metadata protections remain effective. This issue is fixed in versions 1.107.6 and 2.44.0.","status":"published","published":"2026-10-08T15:18:40Z","modified":"2026-10-08T17:28:53Z","cvss":[{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N","level":"low"}],"updated_at":"2026-11-02T17:27:28.108203Z"}],"nvd":[{"id":"CVE-2026-107288","description":"Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.77.0 until 1.107.6 and 2.44.0, the local web_fetch_tool and the WebFetch local fallback compare blocked_domains entries with a URL hostname before both values are normalized to the form used by getaddrinfo. An attacker-influenced model can use an equivalent IDNA spelling, non-ASCII label separator, case variation, or trailing root label that resolves to a blocked host but does not match the configured string, causing the application to fetch that host with its own privileges. allowed_domains fails closed for unmatched spellings, and private-IP and cloud-metadata protections remain effective. This issue is fixed in versions 1.107.6 and 2.44.0.","status":"published","published":"2026-10-08T16:17:04.147Z","modified":"2026-10-13T14:35:05.37Z","cvss":[{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N","level":"low","assessor":"secondary"}],"cwes":["CWE-918","CWE-1289"],"updated_at":"2026-11-02T14:12:55.142936Z"}],"osv":[{"id":"CVE-2026-107288","title":"Pydantic AI: web_fetch_tool blocked_domains bypass via a hostname the resolver normalizes differently","description":"Pydantic AI is a Python agent framework for building applications and workflows with Generative AI. From 1.77.0 until 1.107.6 and 2.44.0, the local web_fetch_tool and the WebFetch local fallback compare blocked_domains entries with a URL hostname before both values are normalized to the form used by getaddrinfo. An attacker-influenced model can use an equivalent IDNA spelling, non-ASCII label separator, case variation, or trailing root label that resolves to a blocked host but does not match the configured string, causing the application to fetch that host with its own privileges. allowed_domains fails closed for unmatched spellings, and private-IP and cloud-metadata protections remain effective. This issue is fixed in versions 1.107.6 and 2.44.0.","status":"published","published":"2026-10-08T15:18:40.047Z","modified":"2026-11-04T02:47:55.900652578Z","cvss":[{"version":"3.1","score":3.7,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N","level":"low"}],"cwes":["CWE-1289","CWE-918"],"updated_at":"2026-11-07T03:07:59.958523Z"}]}}