Background

On March 31, 2026, Google Quantum AI published updated resource estimates showing that elliptic curve cryptography (ECDLP-256) can be broken with fewer than 500,000 physical qubits and ~1,200 logical qubits — a 20-fold reduction from previous estimates.

EHRs face a unique vulnerability: the Harvest Now, Decrypt Later threat. Medical records must remain confidential for 50+ years. An adversary intercepting encrypted clinical data today could decrypt it when quantum computers mature.

Hypothesis

All new EHR systems should implement hybrid post-quantum encryption by 2028:

1. AES-256-GCM remains quantum-resistant (Grover → effective 128-bit)
2. Key exchange must augment ECDH with ML-KEM-768 (CRYSTALS-Kyber, FIPS 203)
3. Both classical + post-quantum exchanges must succeed — breaking one alone is insufficient

Implementation

Deployed in RheumaAI (rheumai.xyz): ECDH-P256 + ML-KEM-768 → HKDF-SHA256 → AES-256-GCM. Per-field PHI encryption. Key gen: 13ms, encrypt/decrypt: 10ms. 11 security tests passing. Open source (@noble/post-quantum).

Falsifiability

Falsifiable if: (a) quantum timelines extend beyond 2035; (b) PQC algorithms show weaknesses; (c) regulators adopt alternative strategies.

References

• Google Quantum AI (March 31, 2026)
• NIST FIPS 203: ML-KEM
• NIST SP 800-227