A cryptography engineer explains why their position on post-quantum cryptography urgency has shifted dramatically. Two recent papers from Google and Oratomic show that breaking 256-bit elliptic curves may require far fewer qubits than previously thought, with Google researchers setting a 2029 deadline. The author argues the risk is now high enough to be 'dispositive' — meaning even a small probability of cryptographically-relevant quantum computers (CRQCs) by 2030 demands immediate action. Concrete recommendations include: shipping ML-DSA signatures now despite their size, treating non-PQ key exchanges as potential active compromises, abandoning hybrid classic+PQ authentication in favor of pure ML-DSA-44, and noting that symmetric encryption (128-bit keys) remains safe. Areas of concern include Trusted Execution Environments (TEEs like Intel SGX and AMD SEV-SNP), ecosystems with cryptographic identities (atproto, cryptocurrencies), and file encryption tools. The author also notes implications for Go standard library cryptography packages.

11m read timeFrom words.filippo.io
Post cover image
Table of contents
Now whatThe picture

Sort: