Bottom line: AI cryptanalysis could either compromise existing algorithms or validate their security through rigorous testing during the post-quantum migration phase.
As the IT industry transitions from RSA and elliptic curves to post-quantum algorithms, a critical window opens for AI cryptanalysis. Cryptographer Matthew Green sees both risk and opportunity for more robust security standards here.
Cryptography is currently in a transition period: traditional public-key schemes like RSA and elliptic curve-based cryptography are being replaced by new post-quantum algorithms. Standards such as HAWK are being considered for this purpose. From a cryptographic perspective, this creates a paradoxical timing problem.
Matthew Green, who advises Anthropic on cryptography, emphasizes: there could not have been a better moment for AI systems to develop cryptanalysis capabilities. The reason is structural: the new post-quantum problems have been less researched than established RSA methods. Should AI systems be able to undermine these still relatively new security assumptions, it would endanger the entire ecosystem’s migration.
However, Green also sees an optimistic outcome: if AI-driven cryptanalysis tools are ultimately unsuccessful against these problems, this creates confidence in the chosen algorithms. Cryptographic literature could simultaneously become more robust, as it is systematically tested with new analytical tools. This assumes, however, that AI systems do not undermine all underlying mathematical problems overall – a scenario corresponding to the theoretical complexity class of Impagliazzo’s “Minicrypt”.
Source: simonwillison.net · Published 29 July 2026
Lumi AI News — AI-assisted curation pursuant to Article 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.7.3.