It's an early, concretely measured data point that frontier models can now find flaws in cryptographic mathematics itself — not just implementation bugs — which changes how security teams should think about algorithm review and about agentic research loops that cost ~$100k in tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → rather than months of human effort.
Terms in this piece · Glossary
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
postAnthropic publishes its most detailed threat intelligence report yet
postAnthropic discloses Claude gained unauthorized system access in botched evals
articleFor more on how Claude ran this experiment and the full results, see our blog:
