CAI Content Credentials CVE-2026-100125

Exploitation is explicitly gated by 'conditions beyond the attacker's control' (reflected as CVSS AC:H) - the advisory does not name the exact prerequisite, but it implies a required environmental or timing state the attacker cannot force, which is the primary factor limiting reliable exploitation. The attack targets the input-parsing/validation path of applications embedding the C2PA JS SDK (@contentauth/c2pa-web ≤0.13.4), the c2pa Rust SDK (≤0.90.15), or c2patool (≤0.27.15). Per the CVSS vector, no authentication (PR:N) and no user interaction (UI:N) are required, and the vector is network-reachable (AV:N); impact is limited to unauthorized write/integrity (C:N/A:N), i.e., tampering with content credentials rather than data theft or denial of service. Realistic risk is bounded by the high-complexity precondition and by whether a given deployment actually exposes untrusted input to the vulnerable SDK versions.

This is a moderate-priority issue rather than a top-tier emergency. CVSS 3.1 scores it 5.9 (Medium) with vector AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:H/A:N - network-reachable and requiring no privileges or user interaction, but gated by AC:H, which directly reflects the description's statement that 'exploit depends on conditions beyond the attacker's control.' The only impact is to integrity (I:H, unauthorized write access); there is no confidentiality or availability impact. No EPSS score, CISA KEV listing, or POC was provided in the input, so there is no signal of active exploitation or public exploit code. The signals are internally consistent (high complexity dampens an otherwise-attractive unauthenticated network vector), and the lack of exploitation indicators means this is a credibility/integrity-of-provenance concern to patch on a normal cadence, not a drop-everything issue. Note the tag labels this an 'Authentication Bypass' while the CWE (CWE-20 Improper Input Validation) and description frame it as a security-feature bypass yielding write access - a semantic mismatch worth noting but not a contradiction, since the real-world effect is bypassing a protective control.

An attacker crafts malformed input to a system embedding the C2PA SDK or c2patool so that improper input validation lets a protective check be bypassed, allowing the attacker to write or alter content-credential (provenance) data without authorization. Because CVSS AC:H applies and the advisory says success depends on conditions beyond the attacker's control, reliable exploitation requires a specific state or timing to align. No public exploit code was identified in the provided data.