Net::SAML2 EUVD-2026-52268

The target Service Provider must be built on Net::SAML2 before 0.86 AND be configured with a decryption key_file so it accepts EncryptedAssertions - this is the exact precondition; SPs with no key_file never decrypt and are not exploitable. Given that configuration, exploitation needs no authentication, no user interaction, and no non-public information: the SP's encryption certificate is deliberately published in its SAML metadata, so any party can encrypt an unsigned assertion to it. The attacker must simply produce a samlp:Response containing an EncryptedAssertion with no dsig:Signature and deliver it to the assertion consumer service endpoint. The only meaningful limiting factor is the key_file/EncryptedAssertion configuration; there is no version-specific, platform-specific, or timing constraint beyond running an unpatched (<0.86) release.

This is a genuine high-priority issue, not a CVSS-inflation case. The provided CVSS 3.1 vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H (9.8 Critical) is well supported by the description: exploitation is network-based, requires no authentication and no user interaction, and yields full identity impersonation. Encryption to a published certificate is intentionally public, so there is no secret the attacker must obtain, and complexity is genuinely low - the attack is deterministic once the SP accepts EncryptedAssertions. The one real limiter, which raw CVSS does not capture, is a configuration precondition: only SPs configured with a decryption key_file (i.e. that actually accept EncryptedAssertions) are exploitable; callers without a key_file never decrypt and are unaffected. No EPSS score, KEV listing, or SSVC decision data was provided, so exploitation-probability and automatability signals are unavailable; the vulnerability is not on CISA KEV and no active exploitation is claimed. The bug was reported by CPANSec (a credible coordinated-disclosure source) and the fixing commit is public, which somewhat lowers the barrier to reverse-engineering a working exploit even though no POC was supplied.

An attacker retrieves the target SP's published SAML metadata, extracts its encryption certificate, and crafts a SAML assertion whose NameID is a privileged victim account. They encrypt this unsigned assertion to the SP's certificate, wrap it in a samlp:Response, and POST it to the assertion consumer service. Because the vulnerable code treats the unsigned decrypted assertion as verified, the SP authenticates the attacker as the chosen user. No POC was provided, but the flaw is deterministic and the fixing commit is public.