RealityHackerOpen in RealityHacker ⇢
Business · OpenAI · published 2026-10-27T00:00:00+00:00 · via TechCrunch

OpenAI and Anthropic Challenge Core Computer Science Theory with AI-Generated Proofs

Image via TechCrunch
Image via TechCrunch

OpenAI claims its AI models have disproven the Exponential Time Hypothesis, a foundational assumption in computational complexity theory, sparking debate among researchers. The company released solutions to three major math problems, with the ETH refutation drawing both excitement and skepticism, including a competing claim from Anthropic. MIT’s Ryan Williams is leading a collaborative effort to vet the findings, emphasizing the need for community scrutiny and new norms around AI’s role in scientific discovery.

Expanded Detail

OpenAI’s claimed refutation of the Exponential Time Hypothesis (ETH) challenges a cornerstone of computational complexity theory, suggesting that certain problems once thought to require exponential time may instead be solvable in sub-exponential time. This would imply that some NP problems could be solved efficiently, potentially undermining the security of widely used cryptographic systems. The claim is supported by AI-generated proofs for three major mathematical problems, with the ETH refutation drawing immediate attention and skepticism from peers.

Context

If validated, the findings could reshape cryptography, algorithm design, and AI development, potentially affecting data security, financial systems, and software integrity. However, the scientific community remains cautious, emphasizing peer review and reproducibility. The broader impact may be a shift in how researchers evaluate AI-generated insights—raising questions about authorship, trust, and the future of collaborative discovery in science.

Expanded detail and Context are AI-generated analysis; the linked article remains the authoritative source.
Read the full article at TechCrunch →
This summary is Al-enhanced to contain extended analysis and broader social context. The original is {NAME); the linked article is the authoritative source. Original headline: “Frontier AI labs say their models refuted the Exponential Time Hypothesis. What does that mean?.” Browse more stories.