R to @OpenAI: We’re releasing the manuscripts, formal Lean certificates, and reasoning walkthroughs…

TL;DR - OpenAI announced results on ten long-standing open problems in mathematics and theoretical computer science, and is releasing the supporting manuscripts, formal Lean proof certificates, and reasoning walkthroughs for public scrutiny. It matters because verifiable, machine-checkable artifacts let mathematicians independently audit AI-derived proofs rather than take them on trust.

• Claimed advances span geometry, cryptography, and complexity theory, per the linked OpenAI index post.
• Release includes three artifact types: human-readable manuscripts, Lean formal certificates, and step-by-step reasoning traces — the Lean certificates are the key verifiability lever.
• Framed as an invitation for the math community to examine and build on the ideas, not just consume the results.
• Content is thin: the post is an announcement teaser, so specific problems, methods, and the model(s) used are not stated here and would need the linked article to confirm.