Over 150 mathematicians around the world have joined forces to warn governments about the hype around artificial intelligence. They say AI’s mathematical abilities are being exaggerated by tech companies chasing investment and market attention. This warning comes as AI tools claim breakthroughs in solving long-standing math problems.

The group released what they call the Leiden Declaration. It stresses that AI models can produce results that look convincing but might be wrong. These flawed outputs could mislead researchers and damage the integrity of mathematics. The declaration urges decision-makers to not rely on flashy announcements or press releases when judging AI’s true capabilities.

Mathematicians fear that AI-generated proofs often lack transparency and proper verification. Unlike human mathematicians, AI systems cannot explain their reasoning clearly, making it hard to check if the proofs are correct. This threatens a core value of mathematics: trust in rigorous, reproducible results.

Pressure on Researchers and the Future of Mathematics

The declaration highlights the growing influence of the tech industry on mathematics research. AI companies offer jobs, funding, and resources that many academics find attractive, especially as university budgets shrink. This creates a tricky situation. Researchers might feel pressured to work with AI firms, even if it conflicts with core scientific values.

Some fear that AI’s rise could steer research toward problems suited to automated tools instead of those that need deep human insight. This could narrow the field and weaken the tradition of free, curiosity-driven inquiry. Early-career mathematicians face particular risks, as their job prospects and funding often depend on embracing AI.

Ethics, Attribution, and Transparency

The declaration points out that many AI models are trained on large collections of published research, often without permission. This raises ethical concerns about credit and copyright. AI-generated work can fail to properly cite the original authors, undermining their contributions.

Mathematicians urge peers to be transparent about how they use AI tools and to keep full responsibility for their results. They want stronger peer review processes adapted to AI-assisted work. Investment in public, open computational resources is also a key demand, to reduce dependence on private AI platforms.

Beyond math, the declaration warns about AI’s broader risks. These include uses in warfare, mass surveillance, and political disruption. Researchers are called to evaluate the ethical consequences of their collaborations and, if necessary, to withdraw from harmful projects.

Leading figures in mathematics emphasize that the field must remain a deeply human endeavor. They argue that human judgment, creativity, and shared community values should guide research, not market pressures or corporate interests. The call is clear: don’t buy the hype. AI is a tool, not a magic wand that replaces human insight.

This pushback comes amid high-profile AI claims. OpenAI recently announced that its model disproved an 80-year-old math conjecture. While impressive, experts note the company released limited details about the model’s workings. This kind of secrecy clashes with the openness needed for scientific validation.

In the end, the mathematicians’ message is a plea for balance. AI can open new doors, but it must be used responsibly. Rigorous standards, transparency, and ethical choices must shape the future of math research. Otherwise, the foundations of the field could erode, with consequences reaching far beyond academia.