An international group of mathematicians from 15 universities across Europe and the US released the Leiden Declaration on June 2, a direct call to protect the core values of mathematics as artificial intelligence reshapes research. The declaration quickly gathered more than 2,500 signatures worldwide, including employees of major AI companies, and is endorsed by the International Mathematical Union.
Mathematicians have taken the lead in confronting AI's impact because the technology is already writing papers, generating "proofs," and assisting in peer review. The goal is not to ban AI but to ensure it strengthens mathematics rather than eroding its foundations. As the declaration's authors put it: "AI offers major opportunities, but without clear choices and shared responsibility, it places the foundations of science under severe pressure."
Five threats to mathematics
The declaration identifies specific ways AI can damage the discipline. AI-generated proofs can appear convincing yet contain nearly invisible errors, overloading the publication system with unreliable results. The technology also raises difficult questions about attribution and copyright because AI systems routinely fail to cite the human work they build on.
A further risk is that mathematicians will soon depend on access to costly proprietary AI tools to produce competitive results, widening inequality between researchers. Press releases and blogs often hype AI claims without scientific scrutiny, fostering overconfidence in systems that embed implicit biases. When commercial interests or technical feasibility steer research agendas, mathematics loses autonomy in setting its own direction, and many areas are drawn into ethical concerns around surveillance, warfare, and environmental damage.
What individual mathematicians can do
The declaration urges researchers to disclose their use of AI, take responsibility for authorship, and put effort into proper attribution. It calls on mathematicians to join public debate, stay informed about emerging technologies, and evaluate the ethical consequences of their work-including withdrawing from harmful projects when necessary. For those needing to stay current, resources like AI for Science & Research provide practical guidance on responsible AI use.
Collective action from institutions
Maintaining rigor, protecting authors' rights, and guaranteeing proper research output will require coordinated effort. Research organizations and publishers should develop clear policies on AI's role. Funding agencies are encouraged to incorporate the declaration's values into their evaluation procedures. Mathematical organizations can facilitate professional codes of practice to strengthen researchers' bargaining position when collaborating with industry.
Beyond mathematics
The same dynamics apply across other academic fields and creative industries. The central question is how to maintain control in a rapidly shifting technological environment-regulating the AI industry and investing in public alternatives so that power does not concentrate in private hands.
Why this matters for science and research
The Leiden Declaration signals that unchecked AI adoption can erode the rigor, attribution, and ethical boundaries that underpin all scholarship. For scientists and researchers in any discipline, staying informed and helping shape institutional policies is no longer optional. Understanding how to deploy AI responsibly-while defending the human judgment at the heart of science-will define the coming decade of research.
Your membership also unlocks:
