OpenAI Release Provides No Evidence Cryptographic Systems Were Broken

OpenAI released mathematical results from an internal frontier model on Oct. 6, with no evidence that Bitcoin (BTC), Ethereum or deployed public-key cryptographic systems have been compromised.

The release includes manuscripts and computer-checkable proof artifacts. Its repository lists 719 manuscripts organized into 372 result families, based on approximately 4,000 problems. About 300 of 719 listed results, or 42%, had been formalized in Lean, a programming language used to check mathematical proofs by computer.

The average result used computing power equivalent to roughly three hours of ChatGPT Pro thinking. That figure describes estimated computing capacity, not literal usage time.

The release has prompted speculation that increasingly capable artificial intelligence systems could eventually expose weaknesses in the mathematical assumptions behind public-key cryptography.

Matthew Green wrote, “I think we might lose public key cryptography.” The statement describes a possible future risk, not a confirmed attack. The release does not announce a cryptanalytic breakthrough, a vulnerability in Bitcoin or Ethereum, or a method for defeating a deployed public-key system.

The repository’s revision history shows that the manuscript collection remains subject to correction. Three manuscripts were withdrawn Oct. 7 after a sign error invalidated an argument involving Weil classes, while 14 other manuscripts were revised.

The project covers a broad range of mathematical results produced by an internal frontier model. The results do not establish an effect on ECDSA, Bitcoin, Ethereum or any deployed public-key system.

For now, the release documents advances in AI-generated mathematical work and an active correction process. The implications for deployed cryptography remain a forward-looking question rather than a confirmed security incident.