Meta updates business terms regarding AI model training on third-party data
Meta has revised its business terms following scrutiny over the use of publisher content for AI training, though legal experts remain skeptical about the scope of the new language.
Meta has adjusted its business terms ahead of an October 30 deadline, removing explicit clauses that appeared to grant the company broad rights to utilize data from news websites for AI model training. The changes follow criticism regarding how the company accesses and processes information hosted on third-party platforms.
Despite the revisions, legal scholars contend that the updated language remains insufficiently restrictive. James Grimmelmann, a professor of digital and information law at Cornell Law School, stated that the remaining phrasing still permits Meta to scrape site content under the guise of supporting its products and services.
This dispute highlights the ongoing tension between large technology companies developing generative AI and digital publishers concerned about intellectual property rights. As legal frameworks surrounding AI training data remain in flux, publishers are increasingly wary of how their content contributes to the development of competing automated systems.
The updated terms will take effect at the end of October. Whether these adjustments will quell concerns from rights holders or lead to further regulatory oversight remains uncertain.