Twenty-five Fields Medallists declare AI benchmark work on open problems a severe misalignment with mathematics

Sources: primary, statement, discussion
Summary: Terence Tao published a statement on his blog, carried in identical text at mathandai.org, signed by 25 Fields Medallists listed with their medal years, that describes the use of open mathematical problems as an AI benchmark as a severe misalignment with the practice of mathematics. The statement names attribution and plagiarism as the concrete harm rather than disputing the capability claims. It invites further endorsements from mathematicians who verify identity through ORCID or an academic email address.
Why it matters: Twenty-five Fields Medallists take a collective position that the labs' push to solve open problems as a benchmark damages the field, and they raise attribution and plagiarism as the concrete mechanism rather than the capability claims themselves.
Follow-up: Whether the signatory list grows past the 25 Fields Medallists and whether any lab answers the attribution complaint.