Meta has announced that it will not sign the new voluntary European Union Code of Practice on the regulation of General Purpose Artificial Intelligence (GPAI) models. About this informs Engadget.
The company explained its decision by the legal uncertainty of the document, as well as criticism due to excessive regulatory interference. Joel Kaplan, Meta’s chief international affairs officer, said the proposed code “goes beyond” the provisions of the AI Act. In his opinion, the implementation of this document may create additional risks for developers and inhibit the innovative development of the industry.
The code, published on July 10, contains a number of requirements for companies. In particular, the use of pirated content during the training of AI models is prohibited, the requests of authors to exclude their works from training data sets are considered, as well as the obligation to regularly update the technical documentation on the functionality of the models.
Although the Code is voluntary, failure to sign it may result in increased scrutiny from regulators. The representative of the European Commission on digital issues, Thomas Rainier, emphasized that companies that do not join the initiative will have to demonstrate compliance with the established requirements in other ways, which may complicate their legal situation.
Violation of the Act on Artificial Intelligence involves financial responsibility – up to 7% of the company’s annual turnover, and for developers of high-risk models – up to 3%.
