Anthropic calls for statutory frontier-AI rules in UK parliamentary evidence. UK policy head Daisy McGregor writes on October 9 that voluntary frameworks are insufficient.
They bind only companies that adopt them. Governments can create industry-wide obligations and international agreements, McGregor argues.
Anthropic cites its proposal already published in June: larger frontier developers should publish safety frameworks, system cards and risk reports, and regularly involve independent evaluators.
McGregor also advocates shared safety standards among democracies. The letter records the company's position and voluntary cooperation; it does not itself implement the proposed duties.