On June 2, the Leiden Declaration on Artificial Intelligence and Mathematics was officially released. It is a landmark document drafted by a working group of 16 researchers. 

The initiative began at the Mechanization and Mathematical Research conference held at the Lorentz Center at Leiden University in September 2025, before being refined through consultation with the international mathematics community. The declaration was formally endorsed by the International Mathematical Union.

It is not a rejection of AI in mathematical research. Rather, it provides a practical framework for responsible AI use in academia. Its central principle is straightforward: use the tool, but take responsibility for the results. Human researchers remain fully accountable for the accuracy of proofs, programming code, citations, and scientific claims. AI cannot be listed as an author.

The declaration identifies three major risks: proofs that appear correct but are ultimately flawed, inadequate documentation resulting from reliance on published work without proper attribution, and the erosion of research independence when scientific priorities are shaped by what AI tools can accomplish rather than by the importance of the underlying questions.

Its recommendations encourage researchers to disclose all AI tools used, assume full responsibility for the accuracy of their work, document original sources carefully, and consider the ethical implications of their research. Institutions, meanwhile, are urged to link funding to these principles, prioritize publication in peer-reviewed journals, and protect authors’ rights as AI systems increasingly rely on copyrighted academic material.

The importance of the declaration lies in its refusal to reduce the debate to the simplistic question of whether AI should be permitted. It accepts that AI will remain part of research. The real question is who is accountable for the outcome. With thousands of mathematicians and researchers from leading institutions supporting the declaration, mathematics has taken an important step toward defining ethical and methodological boundaries for AI-assisted research.

The question now is whether economics needs a similar framework.

Like mathematics, economics relies heavily on modeling, programming, empirical analysis, and statistical inference — all areas where AI can dramatically accelerate research but cannot guarantee accuracy. The danger is that the discipline could become flooded with papers that appear convincing on the surface but are fundamentally flawed. This threatens the credibility of economic research at a time when pressure to publish is increasing and proprietary AI models continue to be trained on copyrighted economic literature without compensation or attribution.

As an economist, I see how easily automated tools can generate sophisticated regressions and models that appear authoritative while producing misleading conclusions. They may confuse correlation with causation, overstate statistical significance, or generate complex outputs without explaining the underlying economic mechanisms. Producing results at the push of a button is not the same as understanding them.

For this reason, I believe national and international economic associations, together with academic institutions across the Arab world and beyond, should adopt standards similar to those established by the mathematics community. Economists should embrace the principle: “Use AI, but own the output.” Publication in peer-reviewed journals and research funding should be tied to adherence to these principles. Full disclosure of AI tools should become standard practice, while human authors must remain solely responsible for the credibility of their analysis, causal claims, and policy recommendations.

In this sense, the Leiden Declaration represents more than a milestone for mathematics. It offers a valuable model for economics and the social sciences as they navigate the opportunities and risks of artificial intelligence. Its strength lies not in restricting technology, but in assigning responsibility where it belongs. It preserves the benefits of AI while protecting the integrity of research.

When researchers place their names on a paper, they affirm that every number, equation, and conclusion is their own responsibility. They receive the credit — but they must also accept the accountability.

• Dr. Abdel-Hameed Nawar is an associate professor of economics at Cairo University.