Parallel research teams publish disproof of long standing math hypothesis

Two leading artificial intelligence labs, Anthropic and OpenAI, have independently derived mathematical proofs that refute a major long standing computer science conjecture known as the Exponential Time Hypothesis. As shared by researcher Levent Alpoge on X, Anthropic reported that an internal experimental Claude model discovered the foundational algorithm autonomously after running through millions of output tokens, with human staff subsequently assisting in organizing the presentation. Both organizations then converted their results into computer verified formats to allow external researchers to audit the logic.

For non experts, this achievement highlights a significant shift in how artificial intelligence contributes to scientific discovery. The Exponential Time Hypothesis has long served as a baseline for understanding the limits of computing power and problem solving efficiency. By using advanced reasoning models to dismantle such a core assumption, and then verifying the math with automated logic tools, researchers are proving that AI systems can solve complex theoretical problems that have challenged human mathematicians for decades.

It remains unclear how quickly this theoretical breakthrough will influence practical applications like code optimization or data encryption standards. Furthermore, the simultaneous discovery by two rival organizations raises an intriguing question: are frontier AI models reaching critical reasoning milestones in tandem due to similar scaling techniques, or did shared academic literature naturally guide both systems to the exact same mathematical shortcut?