RATS Working Group Kayode Oluwole Elekula Internet-Draft Intended status: Informational 16 November 2026 Expires: 20 May 2027 Post-Quantum Cryptographic Considerations for RATS Evidence and Attestation Results draft-elekula-rats-pqc-attestation-00 Abstract This document analyzes post-quantum cryptographic considerations for Remote ATtestation procedureS (RATS) Evidence, Attestation Results, Endorsements, and Reference Values. It describes the distinction between post-quantum key establishment and post-quantum authentication of attestation artifacts; identifies integrity, provenance, freshness, and session-context properties that must remain independently verifiable; and provides implementation guidance for artifact sizing, bounded processing, transformation, transport, failure handling, and privacy. The document does not define a new cryptographic algorithm, attestation token, Evidence format, transport protocol, TLS extension, or key-establishment mechanism. It relies on existing RATS architecture and data models, existing COSE and JOSE mechanisms, and protocol-specific work in the Secure Evidence and Attestation Transport (SEAT) Working Group. The intended result is a set of interoperable considerations that RATS profiles and attestation-system designers can apply when deploying post-quantum-capable attestation systems. Status of This Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at https://www.ietf.org/1id-abstracts.html Elekula Expires 20 May 2027 [Page 1] Internet-Draft PQC for RATS Artifacts November 2026 The list of Internet-Draft Shadow Directories can be accessed at https://www.ietf.org/shadow.html This Internet-Draft is an individual contribution. It does not represent IETF consensus or endorsement by the RATS Working Group. This Internet-Draft will expire on 20 May 2027. Copyright Notice Copyright (c) 2026 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 1.1. Requirements Language . . . . . . . . . . . . . . . . . 3 1.2. Scope and Non-Goals . . . . . . . . . . . . . . . . . . 3 2. Terminology and Security Properties . . . . . . . . . . . . 4 2.1. Properties That Must Not Be Conflated . . . . . . . . . 4 2.2. Key Establishment Versus Artifact Authentication . . . . 5 3. Artifact Protection and Validation Model . . . . . . . . . . 5 4. Algorithm Selection and Cryptographic Agility . . . . . . . 6 5. Artifact Size and Resource Constraints . . . . . . . . . . . 6 6. Transport Considerations . . . . . . . . . . . . . . . . . . 7 7. Session Binding and Context Integrity . . . . . . . . . . . 9 8. Freshness, Replay, and Lifecycle . . . . . . . . . . . . . . 9 9. Evidence Transformation and Provenance . . . . . . . . . . . 10 10. Verifier and Relying Party Processing . . . . . . . . . . . 11 11. Privacy Considerations . . . . . . . . . . . . . . . . . . 12 12. Interoperability and Profile Requirements . . . . . . . . . 13 13. Evaluation and Testing Methodology . . . . . . . . . . . . 14 14. Security Considerations . . . . . . . . . . . . . . . . . . 15 15. IANA Considerations . . . . . . . . . . . . . . . . . . . . 15 16. References . . . . . . . . . . . . . . . . . . . . . . . . 16 16.1. Normative References . . . . . . . . . . . . . . . . . 16 16.2. Informative References . . . . . . . . . . . . . . . . 16 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . 17 Elekula Expires 20 May 2027 [Page 2] Internet-Draft PQC for RATS Artifacts November 2026 1. Introduction Remote attestation allows a Verifier to appraise Evidence about an Attester and produce Attestation Results for use by a Relying Party. RATS defines the architecture and roles for these procedures [RFC9334]. Evidence and Attestation Results can contain cryptographic measurements, claims, endorsements, reference-value identifiers, certificates, signatures, and other data whose size and processing cost vary substantially. Post-quantum cryptography (PQC) changes more than the algorithms used to establish a secure channel. Post-quantum signatures and their public keys can be larger than commonly deployed classical signatures and keys. These differences affect the encoding, storage, transmission, parsing, verification, caching, and lifecycle of attestation artifacts. They can also affect systems that exchange Evidence over constrained links or datagram transports. A system that negotiates a post-quantum or hybrid key-establishment mechanism for TLS does not thereby make its RATS artifacts post-quantum authenticated. Conversely, a post-quantum signature on an artifact does not by itself bind that artifact to a particular TLS or DTLS connection, prove freshness, or establish that the measured platform is currently in an acceptable state. These are distinct security properties and need distinct mechanisms and validation steps. This document provides a framework for identifying and preserving those properties. It is intended to complement, not replace, the RATS architecture, EAT, COSE, JOSE, TLS, DTLS, and SEAT specifications. It does not prescribe a universal maximum artifact size or mandate a single PQC algorithm for every deployment. Profile and local policy remain responsible for selecting acceptable algorithms and resource limits. 1.1. Requirements Language The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals. 1.2. Scope and Non-Goals This document addresses cryptographic and operational considerations for RATS artifacts, including Evidence, Attestation Results, Endorsements, and Reference Values. It discusses interactions with transport security where those interactions affect artifact properties. Elekula Expires 20 May 2027 [Page 3] Internet-Draft PQC for RATS Artifacts November 2026 This document does not define new wire formats, COSE or JOSE algorithms, X.509 algorithm identifiers, TLS or DTLS extensions, handshake messages, exporter derivations, application-layer transport framing, appraisal policies, or attestation semantics. It does not alter the TLS 1.3 or DTLS 1.3 key schedule. Transport-specific binding and exchange procedures are specified by the relevant transport protocol documents, including SEAT work. It also does not claim that every RATS deployment requires post-quantum signatures. Algorithm selection depends on the deployment threat model, applicable policy, cryptographic transition plan, and interoperability requirements. 2. Terminology and Security Properties The RATS terms Attester, Verifier, Relying Party, Evidence, Attestation Results, Endorsements, and Reference Values are used as defined in [RFC9334]. The terms PQC and ML-DSA are used as defined by the relevant cryptographic standards, including [FIPS204] and [RFC9964]. 2.1. Properties That Must Not Be Conflated Artifact integrity: Detection of unauthorized modification to the exact artifact representation or to the authenticated claims it contains. Artifact origin and provenance: Evidence that an artifact, or a particular transformation of it, was produced by an entity authorized for that role. A valid signature alone does not establish that the signer was authorized. Freshness: Evidence that an artifact is sufficiently recent for the relying procedure, using a nonce, timestamp, epoch marker, or another profile-defined freshness mechanism. Session binding: Evidence that an attestation exchange or result is associated with the intended secure-channel session and context. A signature over an artifact does not inherently provide this property. Appraisal: The Verifier's evaluation of Evidence under applicable appraisal policy and trust inputs, and the Relying Party's evaluation of the resulting Attestation Results. Authorization: The decision to permit a requested operation. Successful channel authentication or successful attestation appraisal does not automatically authorize an application operation. Elekula Expires 20 May 2027 [Page 4] Internet-Draft PQC for RATS Artifacts November 2026 2.2. Key Establishment Versus Artifact Authentication Hybrid or post-quantum key establishment protects the establishment of channel traffic keys against the relevant key-recovery threats. It does not change the signature algorithm used to authenticate an Evidence object, Attestation Result, Endorsement, Reference Value, certificate, or detached token component. Each artifact's authentication mechanism must be evaluated independently. Deployments MUST inventory the cryptographic protection applied to each security-relevant artifact. The inventory SHOULD identify the signing or MAC algorithm, key type and size, trust anchor, certificate or key-distribution mechanism, protected representation, and the policy that determines whether the protection is acceptable. A channel's negotiated cipher suite MUST NOT be treated as evidence that every artifact carried over that channel has equivalent cryptographic protection. 3. Artifact Protection and Validation Model A RATS deployment can process multiple independently protected objects. An implementation SHOULD maintain an explicit protection inventory for at least the following artifact classes: * Evidence generated by an Attester or an Evidence-producing environment; * Attestation Results generated by a Verifier; * Endorsements and Reference Values used as appraisal inputs; * Certificates, public keys, trust anchors, and key identifiers used to authenticate those objects; * Detached payloads, manifests, digests, and external references associated with an attestation token; and * Transformation metadata and provenance statements generated when Evidence is normalized, filtered, translated, or otherwise transformed. For each artifact, the receiving party needs to determine what bytes or claims are authenticated, which identity or role is asserted by the signer, whether the signer is trusted for that role, whether the artifact is fresh, and whether it is bound to the current protocol context where required. Validation of one artifact MUST NOT be treated as validation of another artifact merely because the objects arrived in the same transport exchange. Where a format supports detached content, the profile MUST define how the detached content is associated with the protected object. The receiver MUST verify the specified digest or integrity relationship Elekula Expires 20 May 2027 [Page 5] Internet-Draft PQC for RATS Artifacts November 2026 before using the content. Missing, duplicated, ambiguous, or mismatched detached components MUST cause rejection of the affected object. For COSE- or JOSE-based artifacts, profiles SHOULD use registered algorithm identifiers and serialization procedures rather than defining private encodings. [RFC9964] specifies ML-DSA use with COSE and JOSE. A profile selecting ML-DSA MUST state the accepted parameter sets, key representation, algorithm identifier, trust model, and downgrade behavior. 4. Algorithm Selection and Cryptographic Agility Algorithm selection is a profile and policy decision. Implementations MUST NOT infer that an algorithm is acceptable solely because it is syntactically recognized or supported by a cryptographic library. Profiles SHOULD define an explicit allow-list of algorithms and parameter sets for each artifact class. Profiles SHOULD also define how algorithm identifiers are represented, how keys are identified, how certificates or other key-binding mechanisms are validated, and how unsupported or disallowed algorithms are handled. Implementations MUST reject algorithm substitution, ambiguous algorithm identification, and parameter-set confusion. They MUST NOT silently reinterpret an unknown algorithm identifier as a different algorithm. An implementation MUST NOT downgrade to a weaker algorithm merely because verification with a stronger algorithm failed. Any multi-algorithm transition policy MUST be explicit and authenticated by the relevant profile or policy source. Where a deployment accepts both classical and post-quantum signatures during a transition, the policy MUST define whether either signature is sufficient, whether both are required, and which signer identities and trust paths are acceptable. A "hybrid" label alone is insufficient to define these semantics. 5. Artifact Size and Resource Constraints Post-quantum signatures and public keys can increase the size of signed tokens, certificates, and associated metadata. The effect is multiplicative when an artifact contains multiple signatures, certificate chains, endorsements, or nested objects. The resulting size can affect constrained devices, datagram transports, memory-limited Verifiers, and systems with large numbers of Elekula Expires 20 May 2027 [Page 6] Internet-Draft PQC for RATS Artifacts November 2026 concurrent attestation sessions. There is no single maximum artifact size suitable for every RATS deployment. A profile or deployment policy MUST define receiver-side limits appropriate to its transport, implementation capacity, and threat model. Limits SHOULD be specified separately for the complete encoded artifact and for relevant nested components, such as certificate chains, detached payloads, endorsement sets, and transformation metadata. Receivers MUST enforce size limits before allocating memory proportional to attacker-controlled length fields or beginning expensive cryptographic operations. They SHOULD reject an object as soon as its declared or observed size exceeds the applicable limit. They MUST apply limits to both the encoded input and any expanded or decoded representation. Implementations SHOULD bound at least the following resources: * Maximum encoded size of each Evidence and Attestation Result; * Maximum aggregate size of all objects in one attestation exchange; * Maximum certificate-chain length and total certificate bytes; * Maximum nesting depth, collection cardinality, and number of detached components; * Maximum concurrent incomplete objects and aggregate reassembly memory; * Maximum parsing, signature-verification, and appraisal time; * Maximum retries, retransmissions, and total exchange duration; and * Maximum cached object count, cache lifetime, and per-peer cache consumption. Limits SHOULD be configurable by profile and deployment policy. An implementation MUST avoid integer overflow, truncation, unchecked length conversion, unbounded recursion, and allocation based solely on unauthenticated size claims. Resource exhaustion caused by malformed or oversized artifacts MUST NOT cause unrelated sessions to lose their security state or bypass attestation checks. 6. Transport Considerations Transport security protects data in transit according to the selected protocol. It does not remove the need to validate artifact-level signatures, provenance, freshness, or session context. Conversely, artifact-level signatures do not replace transport Elekula Expires 20 May 2027 [Page 7] Internet-Draft PQC for RATS Artifacts November 2026 confidentiality, integrity, peer authentication, or anti-replay properties required by the application. For TLS 1.3 and DTLS 1.3 deployments, protocol-specific mechanisms for conveying and binding attestation exchanges are the responsibility of the applicable transport specification. Exported Authenticators [RFC9261] provide a mechanism for post-handshake authentication context. SEAT documents describe how RATS artifacts can be exchanged using TLS and DTLS-related mechanisms. This document does not define a competing transport or modify those mechanisms. DTLS implementations need to account for datagram size, loss, reordering, retransmission, and handshake fragmentation. An application MUST NOT assume that a large attestation object is delivered atomically in one datagram. Implementations SHOULD avoid dependence on IP fragmentation and SHOULD use the fragmentation and reliability mechanisms defined by the selected transport or application protocol. An implementation SHOULD NOT introduce a second, independent fragmentation and reassembly layer when the selected transport or attestation protocol already provides suitable bounded framing. If a profile requires application-layer chunking, that profile must define: * Unambiguous object identification; * Direction; * Session association; * Total length; * Chunk offset and length; * Duplicate handling; * Overlap handling; * Completion detection; and * Integrity verification. Reassembly MUST be bounded by configured limits and MUST be discarded on timeout, session termination, context mismatch, or validation failure. Chunk boundaries MUST NOT be treated as cryptographic boundaries unless the profile explicitly defines and authenticates them. The receiver MUST verify the complete authenticated object or an authenticated chunking construction before using the object for appraisal or authorization. Elekula Expires 20 May 2027 [Page 8] Internet-Draft PQC for RATS Artifacts November 2026 7. Session Binding and Context Integrity Session binding is distinct from artifact integrity. A valid signature proves that the signed content was authenticated under a key and algorithm; it does not prove that the content was intended for the current connection, peer, role, or transaction. Where an Attestation Result is intended to authorize behavior on a particular secure-channel connection, the transport profile MUST define how the attestation exchange is bound to that connection. The binding input and its interpretation MUST be unambiguous and MUST be authenticated by the mechanism specified in that profile. Implementations MUST NOT invent an exporter label, context encoding, or TLS extension and assume interoperability. The authenticated context SHOULD distinguish, where applicable: * Protocol and version; * Connection or exporter context; * Attesting and relying roles; * Direction; * Requested operation or service; * Attestation exchange identifier; and * Relevant freshness value. The exact fields and cryptographic construction belong to the transport profile. A Verifier or Relying Party MUST reject an Attestation Result when its authenticated context does not match the active exchange. An Attestation Result received on one connection MUST NOT be reused on another connection unless the governing profile explicitly defines a safe transferable authorization model and the result is not represented as connection-specific. When a Verifier transforms Evidence or produces an Attestation Result, the system needs to preserve any session-binding claim required by the relying interaction. A transformation that removes, changes, or fails to carry the relevant context MUST NOT be treated as preserving the original connection binding. 8. Freshness, Replay, and Lifecycle Freshness, signature validity, session binding, and current-state Elekula Expires 20 May 2027 [Page 9] Internet-Draft PQC for RATS Artifacts November 2026 assurance are separate properties. A correctly signed artifact can be stale. A fresh artifact can be bound to the wrong session. A session-bound artifact can describe a state that changed after measurement. Profiles MUST define the freshness semantics required for each artifact class and interaction model. Suitable mechanisms can include unpredictable nonces, authenticated timestamps, or epoch-based mechanisms. RATS architecture discusses freshness and interaction models [RFC9334]. Deployments using epoch mechanisms SHOULD align with the RATS Epoch Markers work rather than define incompatible parallel freshness semantics. When nonces are used, the Verifier or Relying Party MUST ensure that the nonce is: * Unpredictable for the required threat model; * Associated with the intended exchange; * Checked exactly as specified by the profile; and * Not accepted more than once where single-use semantics are required. Timestamp-based freshness requires an explicit acceptance window and a defined response to clock uncertainty. Epoch-based freshness requires validation of the epoch source and the applicable epoch policy. Implementations SHOULD define artifact expiration, cache lifetime, revocation or trust-input update behavior, and invalidation conditions. A cached Attestation Result MUST NOT be reused beyond its defined validity or in a different context unless the profile explicitly permits that use. Expiration of a certificate or signature key and freshness of the measured platform state are separate checks. 9. Evidence Transformation and Provenance RATS systems may transform Evidence to reduce size, normalize formats, remove claims, translate representations, or support intermediary Verifiers. Such transformations can affect both the meaning and the Elekula Expires 20 May 2027 [Page 10] Internet-Draft PQC for RATS Artifacts November 2026 cryptographic protection of the resulting object. Implementations SHOULD follow the semantics and provenance model defined by applicable RATS Evidence Transformation specifications. A transformation MUST NOT be assumed to preserve the original signature over bytes that have changed. The transformed representation needs its own authenticated provenance or a verifiable relationship to the original artifact, as required by the applicable format and profile. Where transformation lineage is security-relevant, the Relying Party SHOULD be able to determine: * The identity or role of the transformer; * The input artifact or its cryptographic digest; * The transformation applied; * The output artifact or its digest; and * Any claims intentionally removed or altered. The exact representation of this information is left to the relevant RATS data model and transformation specification. A transformation MUST NOT silently convert an untrusted artifact into an artifact that appears to have been generated by the original Attester. Intermediaries MUST preserve the distinction between Attester-generated claims, Verifier-generated conclusions, and intermediary-generated metadata. A Relying Party MUST apply trust policy to the entity responsible for each claim it consumes. When a transformation changes a signed object, the receiver MUST validate the protection applicable to the transformed object and MUST NOT infer validity from the signature on the untransformed input alone. Digest references MUST be computed over the exact byte representation specified by the relevant format. 10. Verifier and Relying Party Processing Implementations SHOULD use a staged processing pipeline that rejects invalid input as early as practical: 1. Enforce transport framing, message-size, and exchange-state limits. Elekula Expires 20 May 2027 [Page 11] Internet-Draft PQC for RATS Artifacts November 2026 2. Parse the outer encoding using bounded parsing rules and reject malformed or ambiguous encodings. 3. Check required fields, algorithm identifiers, key identifiers, and profile constraints. 4. Validate cryptographic integrity, signer identity, authorization for the signer role, and trust path. 5. Validate detached content, digests, provenance, freshness, and session context as required. 6. Perform Evidence appraisal and produce Attestation Results according to the applicable RATS procedure. 7. Have the Relying Party evaluate the Attestation Results and its own policy before authorizing the requested operation. This ordering is guidance, not a requirement to expose each internal stage to a remote peer. Implementations MAY reorder operations where necessary, provided they preserve the same security properties and do not perform unbounded or unnecessarily expensive work on unauthenticated input. Implementations MUST distinguish the following conditions in their internal status model: * Transport failure; * Malformed encoding; * Unsupported algorithm; * Size-limit rejection; * Invalid signature; * Untrusted signer; * Stale artifact; * Session-context mismatch; * Failed appraisal; and * Authorization denial. Externally visible errors SHOULD avoid revealing sensitive appraisal details, trust anchors, platform measurements, or policy thresholds to unauthorized parties. Failure to obtain or validate a required attestation artifact MUST NOT be interpreted as successful attestation. If the application requires successful attestation before a protected operation, that operation MUST remain unavailable until the required validation and authorization decisions have completed. 11. Privacy Considerations Post-quantum migration must not become a reason to disclose additional platform identity or measurement data. Elekula Expires 20 May 2027 [Page 12] Internet-Draft PQC for RATS Artifacts November 2026 Profiles SHOULD minimize claims to those required for the Relying Party's stated purpose and SHOULD avoid stable identifiers when unlinkability or pseudonymity is required. Large certificate chains, algorithm inventories, detailed verification errors, and transformation metadata can reveal device type, vendor, software composition, deployment topology, or implementation capabilities. Implementations SHOULD limit disclosure of such information to authorized parties and SHOULD avoid logging complete Evidence or Attestation Results unless operationally necessary and protected by appropriate access controls and retention limits. Algorithm agility and failure reporting SHOULD be designed so that an unauthenticated remote peer cannot use error behavior to enumerate trust anchors, supported hardware, private policy, or detailed platform state. 12. Interoperability and Profile Requirements A deployment profile intended to interoperate across independently developed implementations SHOULD specify: * The RATS interaction model and the roles of each endpoint; * The Evidence and Attestation Result formats and versioning rules; * Accepted COSE or JOSE algorithms, parameter sets, key representations, and trust anchors; * The exact artifact components covered by signatures, MACs, or digests; * Detached-content and transformation-provenance processing; * Freshness semantics, nonce or epoch handling, and replay behavior; * Session-binding requirements and the transport specification that defines them; * Maximum artifact, aggregate, nested-object, and certificate-chain sizes; * Parsing, concurrency, memory, CPU, timeout, and retry limits; * Error handling and the conditions under which application data may be released; and * Privacy requirements, logging policy, and retention periods. Profiles SHOULD include test vectors for successful and unsuccessful verification, including: * Malformed encodings; * Unsupported algorithms; * Invalid signatures; * Mismatched detached content; * Stale freshness values; * Cross-session substitution; Elekula Expires 20 May 2027 [Page 13] Internet-Draft PQC for RATS Artifacts November 2026 * Role confusion; * Oversized objects; and * Transformation-provenance failures. 13. Evaluation and Testing Methodology Implementations and profiles SHOULD be evaluated using reproducible tests that record the artifact format, parameter set, key and signature sizes, platform, cryptographic library and version, transport, network conditions, and resource limits. Performance results without this context are not directly comparable. Measurements SHOULD include at least: * Encoded and decoded sizes for Evidence, Attestation Results, certificates, endorsements, and detached components; * Signature-generation and signature-verification latency; * Parsing and appraisal latency; * Peak memory and CPU utilization per exchange; * Throughput and maximum sustainable concurrent exchanges; * Transport fragmentation, retransmission, and completion time under loss and reordering; and * Time and resource consumption before rejection of malformed, stale, oversized, or unauthenticated input. Latency reporting SHOULD include median and tail percentiles, such as p50, p95, and p99, together with the sample count and test conditions. Tests SHOULD include both normal operation and adversarial cases: * Duplicate and overlapping fragments; * Incomplete reassembly; * Excessive nesting; * Oversized length declarations; * Invalid signatures; * Unknown algorithm identifiers; * Stale nonces or epochs; * Cross-session replay; * Role confusion; and * High-concurrency resource pressure. For DTLS deployments, testing SHOULD include datagram sizes representative of the intended network, packet loss, reordering, duplication, and retransmission. Results SHOULD distinguish cryptographic processing cost from transport-induced delay and application-level queuing. Elekula Expires 20 May 2027 [Page 14] Internet-Draft PQC for RATS Artifacts November 2026 This document does not assert benchmark results. Implementers are expected to publish measured results with enough detail to permit reproduction and comparison. 14. Security Considerations The principal security risk addressed here is the assumption that post-quantum channel security automatically provides post-quantum authenticity for all attestation artifacts. It does not. Each artifact's protection and trust path need independent validation. Large signatures, keys, certificate chains, and nested attestation objects can increase the attack surface for denial of service. Receivers need strict size, parsing, concurrency, memory, CPU, and time limits. These limits must be enforced before attacker-controlled inputs trigger large allocations or expensive cryptographic operations. Replay and cross-session substitution remain possible if freshness and session binding are not independently checked. A signature-valid artifact must not be accepted as fresh or connection-specific without the corresponding authenticated evidence. Algorithm agility creates downgrade and confusion risks. Profiles need explicit algorithm policy and authenticated transition semantics. Implementations must reject ambiguous identifiers and must not silently fall back to weaker algorithms after verification failure. Evidence transformation can undermine provenance if a transformed object is represented as though it were the original signed object. Implementations need to preserve the distinction between original claims, transformed claims, and Verifier conclusions, and validate the applicable provenance for each. Finally, successful cryptographic validation or Evidence appraisal is not equivalent to application authorization. Relying Parties need to evaluate Attestation Results against their own policy and release protected functionality only when all required checks succeed. 15. IANA Considerations This document has no IANA actions. Elekula Expires 20 May 2027 [Page 15] Internet-Draft PQC for RATS Artifacts November 2026 16. References 16.1. Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words", BCP 14, RFC 8174, May 2017. [RFC9334] Birkholz, H., et al., "Remote ATtestation procedureS (RATS) Architecture", RFC 9334, January 2023. [RFC9261] Sullivan, N., "Exported Authenticators in TLS", RFC 9261, July 2022. [RFC9964] Prorock, M. and O. Steele, "ML-DSA for JSON Object Signing and Encryption (JOSE) and CBOR Object Signing and Encryption (COSE)", RFC 9964, May 2026. 16.2. Informative References [FIPS204] National Institute of Standards and Technology, "Module-Lattice-Based Digital Signature Standard", FIPS 204, August 2024. [RFC9147] Rescorla, E., Tschofenig, H., and N. Modadugu, "The Datagram Transport Layer Security (DTLS) Protocol Version 1.3", RFC 9147, April 2022. [RFC9711] Lundblade, L., Mandyam, G., O'Donoghue, J., and C. Wallace, "The Entity Attestation Token (EAT)", RFC 9711, April 2025. [I-D.ietf-rats-epoch-markers] Birkholz, H., et al., "Epoch Markers", draft-ietf-rats-epoch-markers-08, November 2026, Work in Progress. [I-D.ietf-rats-evidence-trans] "Evidence Transformations", draft-ietf-rats-evidence-trans-03, October 2026, Work in Progress. [I-D.fossati-seat-expat] "Remote Attestation with Exported Authenticators", draft-fossati-seat-expat-05, November 2026, Work in Progress. [I-D.reddy-seat-expat-transport] Elekula Expires 20 May 2027 [Page 16] Internet-Draft PQC for RATS Artifacts November 2026 Reddy, T. and H. Tschofenig, "Application-Layer Transport for Exported Authenticators and Attestation", draft-reddy-seat-expat-transport-02, August 2026, Work in Progress. Author's Address Kayode Oluwole Elekula Lagos Nigeria Email: kayodeelekula@gmail.com Elekula Expires 20 May 2027 [Page 17]