Post-quantum readiness for AI systems

What it is for

AI systems move and keep data that stays sensitive for years: training data, embeddings, prompts and logs, model weights, and the signatures that prove a model is the one you approved. Data encrypted today with quantum-vulnerable algorithms can be recorded now and decrypted later. G3 puts post-quantum migration into the same plan as the AI system.

The timelines

• NIST published the first post-quantum standards on 13 August 2024: FIPS 203 (ML-KEM) for key establishment, and FIPS 204 (ML-DSA) and FIPS 205 (SLH-DSA) for signatures.^⁠[1][2][3]
• NIST IR 8547 became final on 27 October 2026.^⁠[4] It deprecates quantum-vulnerable public-key algorithms such as RSA and elliptic-curve schemes after 2030 and disallows them after 2035.^⁠[5]
• The EU roadmap asks Member States to begin the transition by the end of 2026 and to move critical infrastructure by the end of 2030 at the latest.^⁠[6]
• The UK’s NCSC sets milestones for 2028, 2031 and 2035, with migration of all systems, services and products complete by 2035.^⁠[7]

What we look at

• TLS and key exchange between users, applications, model providers, vector stores and tools.
• Signatures on models, packages, containers and attestations.
• Key management, certificates and the libraries that implement them.
• Data that must stay confidential beyond 2030.

What you receive

A cryptography inventory in CycloneDX’s cryptography bill of materials format,^⁠[8] and a migration plan that says what to change first and by when.