OpenAI says its models have produced proofs tackling local smoothing in every dimension, refuting the Exponential Time Hypothesis, and establishing the existence of certain four-dimensional quantum field theories. If upheld, these would redraw parts of mathematics, algorithms and theoretical physics. You asked AI to tidy your inbox. It has apparently decided computational complexity needs tidying first. The receipts are uneven. Math and computer-science writeups and Lean formalizations are public; the physics artifacts are being held for an external group’s writeup. OpenAI calls the manuscripts raw research data, and expert groups are examining what they mean. This is not an independently settled three-for-three scorecard or a finished quantum computer in your shopping basket. The company says collaborators participate voluntarily and unpaid. So the machine gets the breakthrough headline, and the humans get a group project. At least this time someone actually brought the working.