AWS-LC CVE-2026-3338

Exploitation requires the target application to invoke AWS-LC's PKCS7_verify() on attacker-supplied PKCS#7/CMS objects AND to rely on the boolean verification result for an integrity or authentication decision; the specific trigger is a PKCS7 object that includes Authenticated Attributes (signedAttrs), which is the code path where validation is improper. No authentication is needed (CVSS PR:N) and no user interaction (UI:N), and the attack is network-deliverable wherever such a signed blob can be submitted. Limiting factors: applications that never verify PKCS#7 messages, never accept ones containing Authenticated Attributes, or do not trust the result for a security decision are not meaningfully exposed; impact is confined to integrity (no data disclosure or denial of service); and the component must be an unpatched AWS-LC build (pre-1.69.0), including the aws-lc-sys/aws_libcrypto distributions.

Signals are mixed but lean toward prioritize-the-patch rather than emergency. The CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:N/UI:N) describes a network-reachable, low-complexity, unauthenticated attack with no user interaction and High vulnerable-system Integrity (VC:N/VI:H/VA:N), yielding 8.7 - appropriate for a signature-validation bypass whose entire consequence is integrity, not confidentiality or availability. Against that, EPSS is very low (0.03%, 7th percentile), there is no CISA KEV listing and no public exploit identified, indicating no observed in-the-wild activity. The key caveat is contextual exploitability: although the vector is marked network/unauthenticated, real impact only materializes in applications that actually call PKCS7_verify() on attacker-influenced PKCS#7 objects and trust the result for an authentication or integrity decision - a deployment that never verifies PKCS#7 with Authenticated Attributes is not meaningfully exposed. Net: a genuine cryptographic-integrity flaw worth patching promptly, but the absence of KEV/POC/EPSS pressure suggests targeted-not-widespread risk.

An attacker crafts a PKCS#7 signed object that includes Authenticated Attributes but whose signature does not legitimately cover the claimed content, then submits it to an application that calls AWS-LC's PKCS7_verify() to authenticate the message. Because verification is improperly performed, the forged object is accepted as validly signed, letting the attacker impersonate a trusted signer or smuggle unauthorized content - for example bypassing a signature-gated authentication or token-validation step. No authentication or user interaction is required (AV:N/AC:L/PR:N/UI:N), and since no public exploit is currently known this would be attacker-developed.