RUSTSEC-2026-0585 json
ML-KEM does not perform the FIPS 203 input checks
published yesterday modified yesterday
- cvss
- —
- not scored
- cisa kev
- Not listed
- not in CISA's catalog
- epss
- —
- no score, EPSS covers CVEs only
- public exploits
- 0
- none known
Summary
`pqcrypto-mlkem` wraps PQClean's ML-KEM implementation, which does not perform the input checks required by FIPS 203:
- encapsulation does not perform the encapsulation key modulus check (section 7.2), so encapsulation keys with coefficients that are not reduced modulo q are accepted; - decapsulation does not perform the decapsulation key hash check (section 7.3), so a decapsulation key whose stored hash or embedded encapsulation key has been altered is accepted.
With the Wycheproof ML-KEM vectors, 108 / 112 / 116 invalid encapsulation keys (ML-KEM-512 / -768 / -1024) and 2 invalid decapsulation keys per parameter set are accepted.
This does not disclose key material. It is a FIPS 203 non-conformance, and an affected implementation can disagree with a conforming one about which keys are valid.
The crate is unmaintained (see RUSTSEC-2026-0161) and no patched version is expected. Migrating to a maintained implementation, such as the [`ml-kem`](https://crates.io/crates/ml-kem) crate, is recommended.
Is RUSTSEC-2026-0585 in your dependencies? Fencer shows which of your repositories are affected as soon as an advisory lands, and opens the fix as a pull request.
Check with FencerAffected packages
| ecosystem | package | affected | fixed in |
|---|---|---|---|
| crates.io | pqcrypto-mlkem | >=0.0.0-0 |
no fix listed |
References
Sources
Fencer Intel keeps one entry per vulnerability: the CVE id wins, then GHSA, and records that alias each other are merged, except that two CVEs always keep separate entries. The same data is available as JSON.