MANILA, Philippines – Nearly 2,000 mathematicians from renowned universities and academic institutions around the world including Harvard, Columbia University, University of Cambridge, and University of Oxford, signed the Leiden Declaration on Artificial Intelligence and Mathematics in June 2026, calling for oversight and ethical use of the technology in math research. 

The declaration establishes an ethical framework for mathematicians, asserting, above all, the primacy of human autonomy in their research. 

The declaration warns, “[AI] developments put the autonomy of mathematics under threat. The increasing involvement of technology companies in mathematical research raises the risk that research questions may come to be prioritized because of their amenability to automated mathematics, rather than expert judgment of their deeper significance.” 

In other words, it’s the human that should set the agenda for what is important research rather than choosing such because a topic plays well with automated AI tools. 

The declaration brings up its issues with AI: Its tendency to produce “plausible but unreliable (or even incorrect)” arguments; improper attribution and the use of copyrighted work in their data corpus; and the “overemphasizing” of the technology’s contribution to the research work. 

To deal with these, first and foremost, it insists on the primacy of tool disclosure, requiring authors to include a dedicated section in their papers detailing the use of all automated systems, and large language models. 

It asks researchers to ensure proper attribution of the human labor behind any synthesized result, and ensure that the responsibility for the correctness and adequacy of any scientific argument will remain exclusively with the human authors, who must personally verify every claim regardless of how it was generated.

This call for accountability and the establishment of an ethical AI framework follows similar ones from other fields such as medicine, education, and journalism. It comes amid a recent study in May that found 146,900 AI-generated fake citations, as reported by CNET, appearing in research papers across major databases like arXiv and PubMed Central. This “hallucinating with confidence” threatens to clutter the literature with non-existent references and incorrect results, potentially misdirecting future researchers who build upon these faulty foundations. 

The declaration also makes an urgent call for “significantly increased” government regulation and public oversight of the technology industry, noting the involvement of AI in military programs, mass surveillance, the development of technologies that promote misinformation or undermine democratic processes, and its environmental impact as well. 

It urges governments to provide stronger protection to copyrighted work, and to establish public computational infrastructure at university, national, and international levels. By funding public alternatives to proprietary technologies, specifically computer clusters for mathematical modeling and machine learning, policymakers can ensure that the tools of scientific discovery remain transparent and independent of private corporate control.

Many AI tools today are in the hands of large private tech firms. In the US, Senator Bernie Sanders offers a radical idea, arguing that the public should have a 50% stake in AI companies via his proposed American AI Sovereign Wealth Fund Act. 

The mathematicians’ demand for public oversight, aside from academic and research integrity, also stems from concerns regarding the power of commercialized AI entities that are currently incentivized to overhype their products. 

In forming policies, the declaration says: “There is currently a strong commercial incentive on the part of the technology industry to overstate the capabilities of their products. Consult with experts, including mathematicians, in forming policy decisions rather than relying on press releases or popular reporting of mathematical results.” 

Among the signatories, Leslie Ann Goldberg, head of computer science at University of Oxford, said, “…research in mathematics (and in mathematical disciplines like theoretical Computer Science) almost always builds on previous research, so it is essential for researchers to know that the results in the literature are correct.”

“Inaccurate AI-generated drafts are cheap to produce, and there is a risk of cluttering the literature with claimed results that are simply wrong. Once that happens, the errors are likely to propagate as new results are built on faulty foundations. I welcome the recommendations of the Leiden Declaration, particularly the disclosure of tool use, and continued publication through peer-reviewed journals,” she said. – Rappler.com