[Seat] Re: [Ufmrg] Re: Re: Comments on formal analysis of relay attacks in attested TLS (CVE-2026-33697)

[Dr Küçük Oxford University DPhil Computer S cience <[e-mail]>]

Dear all,
I have reviewed the paper intra-handshake.fail, the accompanying draft (draft-intra-handshake-fail-01), and the artifacts of the formal analysis of relay attacks in attested TLS at commit 8ba77c5. I would like to strongly support this work and share a few comments.

First, this paper has genuinely moved the field forward. The formal treatment of relay attacks in attested TLS gives the community a much stronger foundation for reasoning about these designs. The vulnerabilities existed for several years, yet manual testing, intuition, and audits had not discovered them.

Second, the practical impact is significant: all vendors of intra-handshake attestation have acknowledged the attacks and have either published security advisories or archived the affected repositories (see Section 4: https://www.ietf.org/archive/id/draft-intra-handshake-fail-01.html#section-4) In addition, the vulnerable draft draft-fossati-tls-attestation has been withdrawn by its authors (see abstract and introduction: https://datatracker.ietf.org/doc/draft-fossati-tls-attestation/10/)

Third, the severity assessment speaks for itself — this is the highest-scored CVE in the history of Confidential Computing (see Section 5: https://www.ietf.org/archive/id/draft-intra-handshake-fail-01.html#section-5)

Fourth, regarding the concerns raised by Nathanael in the recent thread (https://mailarchive.ietf.org/arch/msg/seat/aEV9dUFotAQzHndk23qBcwBT3as/) I found the authors' clarifications convincing and I support their responses. In my reading, the concerns do not identify a concrete gap in the formal analysis. Given the substantial body of work already established by Sardar et al. over the past decade, I believe the working group is better served by building on that foundation than by developing a parallel approach from scratch, and I would encourage collaboration toward that end.

Finally, a question for the authors of the hybrid proposals (draft-fossati-seat-early-attestation and draft-ritz-seat-facts): which specific security property, as modeled in ProVerif, can the hybrid proposals achieve that post-handshake attestation alone cannot? A precise statement of this distinguishing property would help the working group evaluate whether the added complexity of the hybrid approach is justified.

Thank you to the paper authors for this important contribution and for promptly addressing working group feedback.

Best regards,

Dr Küçük

DPhil Oxford University
https://www.cs.ox.ac.uk/people/dr.kucuk/
x-ETH Zürich
[e-mail]