Frontier AI labs say their models refuted the Exponential Time Hypothesis. What does that mean?

Researchers at OpenAI have claimed that their AI models discovered a mathematical proof suggesting the Exponential Time Hypothesis is incorrect. This hypothesis is a foundational assumption in computer science that sets limits on how quickly certain complex problems can be solved. If the findings are verified, it could fundamentally reshape computational complexity theory by suggesting that some problems once thought to be intractable may be solvable more efficiently than previously believed.