Content Credentials CVE-2026-48296

The vulnerable application must invoke the C2PA parsing API on attacker-controlled input - a crafted C2PA manifest embedded in a media file (image, video, audio, or document). No authentication to the target application is required by the flaw itself (PR:N), and no user interaction beyond supplying the file is needed (UI:N). Exploitation is limited to scenarios where the application actually processes C2PA manifests from untrusted sources; applications that only generate or sign manifests, or that only process manifests from fully trusted internal sources, are not meaningfully exposed. The local attack vector (AV:L) means direct remote exploitation against the SDK in isolation is not applicable, but indirect exposure arises when c2pa-web or c2patool are embedded in network-facing services that accept file uploads or URLs pointing to signed content. Impact is confined to availability - no data exfiltration or code execution pathway is identified.

Real-world risk is moderate-to-low. The CVSS base score of 6.2 is driven by AV:L/AC:L/PR:N/UI:N, which correctly captures that no privileges or user interaction are needed to trigger the crash once a malicious manifest is supplied to the parser - but the local attack vector limits the exposure surface compared to a direct network-exploitable flaw. EPSS at 0.15% (5th percentile) and SSVC 'exploitation: none / automatable: no' are mutually reinforcing: no in-the-wild or POC exploitation has been observed, and the attack cannot be trivially mass-automated across internet-facing endpoints. The technical impact is rated 'partial' by SSVC, consistent with an availability-only (A:H, C:N, I:N) outcome and unchanged scope. KEV status is absent. The primary real-world concern is for pipelines that ingest C2PA-signed content from untrusted third parties (e.g., media authentication services, content moderation workflows, camera firmware update chains) - in those contexts the 'local' vector effectively becomes network-reachable. A key signal gap: no vendor advisory text is directly quoted here beyond the APSB26-80 reference URL, so root-cause details (exact field triggering underflow, stack trace, crash type) are inferred from CWE-191 and description rather than confirmed from source code or advisory body.

An attacker with the ability to supply a crafted C2PA-signed media file to a target system - for example, by uploading an image to a content authenticity verification service, submitting to a digital asset management system, or placing a file in a monitored directory - causes the Content Credentials parser to process a manipulated length or count field. The integer underflow produces a wrapped value, triggering a panic or memory fault that crashes the consuming application. No public proof-of-concept code has been identified at time of analysis.