Axiom Math wants to keep AI slop out of mathematics
News Source
β’Thu, 05 Nov 2026 16:16:45 +0000
π° What Happened
Axiom Math, a startup valued at 1.6 billion dollars, says it has formally verified the Stacks Project. Stacks is one of the largest bodies of mathematics ever written by people, covering algebra, geometry and number theory. Its roughly 7,654 pages were turned into about 16 million lines of computer-checked code.
A fleet of AI agents did the work, translating proofs into a language called Lean. A proof assistant then checks the code line by line.
π The Backstory
The Stacks Project was started in 2005 by Columbia mathematician Johan de Jong as a free, open reference. Mathematicians have added to it ever since.
Formal checking means no step can be skipped or faked, because a program tests every one. Axiom Math wants to build a trusted base that AI agents can use to write software and, later, do science. The announcement follows a flood of AI-written maths papers, including 719 manuscripts OpenAI released last month.
π― Why It Matters
If AI can check the maths it writes, the results become trustworthy, which matters for medicine, engineering and finance.
Axiom Math, a startup valued at 1.6 billion dollars, says it has formally verified the Stacks Project. Stacks is one of the largest bodies of mathematics ever written by people, covering algebra, geometry and number theory. Its roughly 7,654 pages were turned into about 16 million lines of computer-checked code.
A fleet of AI agents did the work, translating proofs into a language called Lean. A proof assistant then checks the code line by line.
The Stacks Project was started in 2005 by Columbia mathematician Johan de Jong as a free, open reference. Mathematicians have added to it ever since.
Formal checking means no step can be skipped or faked, because a program tests every one. Axiom Math wants to build a trusted base that AI agents can use to write software and, later, do science. The announcement follows a flood of AI-written maths papers, including 719 manuscripts OpenAI released last month.
If AI can check the maths it writes, the results become trustworthy, which matters for medicine, engineering and finance.