AI is rewriting math. Can we trust its proofs?

Artificial intelligence is changing how mathematicians work. A growing number of AI systems can now suggest proofs and solve problems once handled only by humans. But a new report in *Science* (Volume 394, Issue 6823, October 2026) asks a difficult question: can we trust the results? A mathematical proof is a step-by-step argument that shows a statement is true. Traditionally, human experts check each proof by reading it carefully. AI-generated proofs can be long and complex, making this check harder. Some systems produce answers that look correct but contain hidden errors. Researchers are now testing ways to verify AI proofs automatically. These tools check every logical step, much like a spell-checker for mathematics. The goal is to confirm that a proof is valid before anyone relies on it. The debate matters beyond the classroom. Mathematicians use proofs to support work in computer security, engineering, and science. A wrong proof can lead to costly mistakes. For now, experts agree on one point: AI is a powerful assistant, not a replacement for human judgment. The challenge is to combine speed with accuracy, and to keep trust in the results.