Mathematicians Issue Leiden Declaration Warning on AI Risks
A group of 16 prominent mathematicians from 15 universities across eight countries has issued the Leiden Declaration on Artificial Intelligence and Mathematics, an 11-page document warning that the rapid rise of AI poses fundamental threats to the core values of mathematical research. Published on June 2, 2026, and officially endorsed by the International Mathematical Union (IMU), the declaration calls for responsible development, transparent disclosure of AI tool use, and regulatory safeguards to protect the integrity of the discipline.
A Community Responds to Urgent Challenges
The declaration emerged from a September 2025 workshop at the Lorentz Center at Leiden University, titled “Mechanization and Mathematical Research,” which brought together approximately 60 participants from 10 countries — including mathematicians, computer scientists, philosophers, and social scientists. Over the following eight months, a working group of 16 researchers developed the declaration with extensive feedback from the global mathematical community, as Leiden University reported.
As of June 4, 2026, the declaration had garnered over 1,213 signatories, including two Fields Medal laureates: Peter Scholze, Director of the Max Planck Institute for Mathematics, and Terence Tao of UCLA. Both have publicly endorsed the document, with Scholze calling it “a wonderful declaration, coming at the right time.”
The Five Threats AI Poses to Mathematics
According to the official declaration website, the document identifies five specific threats that unchecked AI development poses to the mathematical discipline:
1. Unreliable Results: AI can produce plausible but incorrect proofs that are difficult to distinguish from correct ones, putting the peer review system under severe pressure. Leslie Ann Goldberg, Head of Computer Science at the University of Oxford, warned that “inaccurate AI-generated drafts are cheap to produce, and there is a risk of cluttering the literature with claimed results that are simply wrong.”
2. Lack of Attribution and Copyright Violation: AI models trained on published mathematical works often fail to properly cite sources, undermining the traditional system of attribution and intellectual property.
3. Dependence and Inequality: Researchers with access to expensive proprietary AI and computing resources gain unfair advantages, marginalizing those without such access or those unwilling to use commercial tools.
4. Overhyping of Results: Results communicated through press releases and blogs bypass peer review, overstating AI capabilities and undervaluing human contributions.
5. Loss of Autonomy: Commercial interests may steer research toward problems amenable to automation rather than those of deeper mathematical significance, eroding the discipline’s autonomy.
A Catalyst for Urgency
The declaration’s urgency was amplified by a recent breakthrough: in May 2026, OpenAI reported that its latest general reasoning model independently cracked an 80-year-old mathematical conjecture, as The Paper (China Science Daily) reported. This event highlighted just how rapidly AI is advancing in mathematics — from solving simple trick problems to independently cracking long-standing open questions.
What the Declaration Calls For
The Leiden Declaration does not call for banning AI from mathematics. Instead, it urges clear community norms for its use. Key recommendations include:
- For individual mathematicians: Disclose all AI tool use transparently in papers, retain full human responsibility for correctness, and actively participate in public discourse about AI.
- For organizations and funders: Develop clear policies for AI use in publishing and reviewing, protect authors’ rights to opt out of training data use, and support public research laboratories independent of industry.
- For policymakers: Strengthen legal protections for authors, regulate the AI industry, and invest in public computational infrastructure.
- For commercial AI companies: Abide by the same standards expected of academic research and respect researchers’ freedom of conscience.
Ulrike Tillmann, IMU Vice President, stated: “By endorsing the declaration, the IMU affirms that the future of mathematical research must be guided by human judgment, fair and transparent practices, and the shared values of the global mathematical community. Mathematics is, and should always remain, a profoundly human endeavour.”
Broader Implications
The declaration aligns with other international initiatives including the Uppsala Code of Ethics for Scientists and the UNESCO Recommendation on Open Science. It also arrives amid a broader policy landscape where governments are grappling with AI’s rapid advancement — including a recent U.S. executive order granting federal officials priority access to advanced AI models.
What’s Next
The declaration will be discussed at a dedicated forum during the International Congress of Mathematicians in Philadelphia in July 2026, a critical venue for further debate and potential action. As IT Home noted, the declaration raises fundamental questions about whether the mathematical community can maintain its autonomy as tech companies invest heavily in AI-driven mathematics.
Peter Scholze captured the human dimension of the crisis: “The goal of mathematical research is human understanding of mathematics, and so mathematics can only thrive in a community of human mathematicians. It is crucial to preserve this communal spirit.”