• chain-of-thought — Having a model write out intermediate reasoning steps before its answer, which markedly improves performance on hard problems.

A frontier lab publishing model-assisted cryptanalysis against HAWK and AES, with the reasoning traces released, is a concrete data point on where AI now sits in offensive-security research.

• articleFor more on how Claude ran this experiment and the full results, see our blog:
• articleAs part of our Responsible Scaling Policy, we publish regular Risk Reports. Thes
• articleDiscovering Cryptographic Weaknesses with Claude
• postClaude designed working protein binders for 14 of 15 targets

• articleDiscovering Cryptographic Weaknesses with ClaudeAnthropicAI
• articleCryptanalysisBench: Can LLMs do Cryptanalysis?AnthropicAI
• articleStealing Reasoning Traces from Proprietary LLM APIssimonwillison.net

Checking sign-in…

Loading comments…