Azure AI Foundry EUVD-2026-82594

No special conditions - remote unauthenticated exploitation over the network against Azure AI Foundry (AV:N/AC:L/PR:N/UI:N). Exploitation requires the attacker to reach an Azure AI Foundry feature that fetches an attacker-influenced URL server-side. The primary limiting factor is that this is a Microsoft-operated cloud service: the fix is applied server-side by Microsoft with no customer action, so the exposure window is bounded by Microsoft's remediation timeline rather than by customer patching, and impact is confined to confidentiality (no integrity or availability effect, no scope change to other systems).

The vendor CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N, 7.5 High) describes a remotely reachable, low-complexity, unauthenticated SSRF whose only impact is high confidentiality loss - integrity and availability are both N and scope is Unchanged. Note a labeling conflict: the description says the flaw lets an attacker 'elevate privileges,' but the CVSS impact metrics (C:H/I:N/A:N, S:U) describe pure information disclosure, not privilege elevation - this is Microsoft MSRC boilerplate phrasing and the vector, not the prose, should drive prioritization. No EPSS score, no CISA KEV listing, and no public exploit are present in the input: exploitation status is 'no public exploit identified at time of analysis.' There is no SSVC decision data supplied. Because Azure AI Foundry is a Microsoft-operated cloud service, the fix is almost certainly server-side and already applied for all tenants, which materially lowers real-world customer action risk even though the raw score is High. Net: a genuine but self-remediating SSRF-driven disclosure issue - real severity, low residual customer exposure once Microsoft ships the fix.

An unauthenticated remote attacker sends a crafted request to an Azure AI Foundry endpoint that fetches a user-supplied URL, coercing the service into making server-side requests to internal Azure resources or the instance metadata endpoint and returning the response, disclosing sensitive internal data. No authentication or user interaction is required (PR:N/UI:N) and complexity is low (AC:L). No public proof-of-concept is identified at time of analysis.