IETF last calls and TLS list disputes

IETF activity centered on post-quantum TLS, with a heated Last Call on pure ML-KEM key agreement and a parallel dispute over how the TLS working group judges consensus. Separate threads advanced Keccak-based ciphers, LLM handling in standards work, and several smaller protocol proposals.

ML-KEM TLS 1.3 draft enters Last Call amid security objections

The IETF has issued a Last Call for draft-ietf-tls-mlkem-09, which specifies ML-KEM post-quantum key agreement for TLS 1.3 as an Informational RFC. Participants oppose the pure (non-hybrid) design on security grounds and question whether rough consensus exists. The outcome will shape whether TLS deployments can adopt standalone lattice KEMs without classical fallbacks.

TLS chairs' consensus call on ML-KEM draws transparency challenge

A related thread on the TLS list contests the working-group chairs' rough-consensus determination for the ML-KEM RFC. Critics object to undisclosed weighting of designated expert participants and demand greater process transparency. The dispute directly affects how future cryptographic consensus is recorded on the list.

Keccak XOF ciphers proposed for TLS 1.3 key schedule

A new draft, draft-sullivan-tls-xof-ciphers-00, proposes replacing parts of the TLS 1.3 key schedule with Keccak-based extendable-output functions. Members of the Keccak team and volunteers offering formal analysis have joined the discussion of optimizations. The work would give TLS an alternative to the current HKDF construction rooted in SHA-2.

Draft on handling LLM-generated text in IETF discussions

An IETF-wide thread examines a draft that would set norms for the use of large-language-model output in working-group mailing lists and documents. Participants are debating disclosure requirements and acceptable assistance levels. Clear rules would affect how contributions are evaluated across every IETF area.

DMARC working group weighs making ARC Historic

The DMARC working group is evaluating consensus to reclassify ARC (RFC 8617) as Historic and to publish a report on the experiment. Discussion focuses on whether the authenticated-received-chain mechanism has failed to gain sufficient deployment. A Historic status would signal that operators should not rely on ARC for new authentication paths.

OpenPGP card fingerprint encoding for v6 keys

OpenPGP participants are examining an encoding scheme that fits 20-byte card fingerprint fields so existing hardware security devices can hold v6 keys. The thread seeks a practical mapping that avoids breaking deployed OpenPGP cards. Compatibility here determines whether users can move to the newer key format without replacing tokens.

Formal analysis of attested-TLS relay attacks noted

A message on the SEAT list endorses formal analysis of relay attacks against attested TLS, referenced as CVE-2026-33697, and urges the group to take up the findings. The single contribution highlights a concrete vulnerability class in attestation-bearing TLS handshakes. Attention from SEAT would determine whether protocol changes follow.

AIPREF considers AI System Inference and AI User Input categories

The AIPREF working group is reviewing a pull request that would add "AI System Inference" and "AI User Input" categories to its vocabulary. Debates center on precise definitions and the intended scope of each label. Acceptance would expand the set of control signals available for AI-related data flows.