OpenAI and Anthropic claim to refute Exponential Time Hypothesis

OpenAI and Anthropic have jointly claimed to refute the Exponential Time Hypothesis (ETH), a fundamental assumption in computer science stating that certain logic puzzles require exponential time to solve. OpenAI initially released solutions to three major problems, prompting immediate skepticism from the scientific community regarding its 'proof dump' methodology. Anthropic subsequently published its own refutation of ETH, lending credibility to the claim that AI models can find sub-exponential algorithms for these tasks. MIT professor Ryan Williams is organizing a collaborative response to verify the proofs and assess potential impacts on cryptographic security, noting that while current findings don't break existing encryption, they challenge long-held intuitions about computational hardness.