Study estimates cryptanalysis effort across public-key algorithms using AI-assisted pipeline

A new analysis attempts to quantify how much cryptanalytic effort has been directed at major public-key algorithms by mining academic literature. The work, led by cryptographer Matthew Green, classifies roughly sixteen thousand candidate papers on topics including the Elliptic Curve Discrete Logarithm Problem (ECDLP), structured and plain lattices, and syndrome decoding, then converts these publications into estimated research-allocation years using a person-level author map.

The study reports that lattices have overtaken elliptic curves in cumulative cryptanalysis attention only recently, framing this history as "inference budgets" under stated productivity assumptions. Green emphasizes that the estimates are derived from published work and may substantially overstate the actual time spent by researchers, since unpublished cryptanalysis—including classified intelligence-agency work—is not captured.

The analysis pipeline and written output were produced by AI models—specifically Claude Fable 5.1, Claude Opus 5.5, and GPT-6 Astra—while Green set the research questions, made key methodological decisions, and oversaw the project. Green notes that the approach relies on published literature as its instrument, which excludes both unpublished academic results and secret state-level cryptanalysis.

The author highlights several caveats: the method cannot account for unpublished or classified work, and the conversion from papers to person-years depends on assumptions about individual productivity. Despite these limitations, Green describes the effort as a "good first stab" at creating a measurable proxy for cryptanalytic attention over time.