Meta backtracks on wording that raised concerns about using website content for AI training
Amman3 min read
Share
Proposed amendments to Meta's commercial terms of use sparked controversy over the potential use of website content in developing artificial intelligence technologies, before the company backtracked on wording that had raised legal and media concerns regarding the scope of its authority to collect and use data.
The issue began on October 15, after "Nieman Lab," which specializes in journalism and the future of media, revealed new clauses in Meta's commercial terms that were scheduled to take effect on October 30.
The original wording included allowing the company to use automated tools, including web crawlers, to access websites, pages, and links associated with the products and services of commercial account holders, and to periodically collect, index, analyze, and update the data contained therein.
The controversy focused on a clause that indicated the possibility of using the content collected by the company to provide, develop, and improve its products, as well as to conduct research, train artificial intelligence systems, and enhance the safety of services.
This text raised concerns that data collection operations might extend to articles and reports published on the websites of media organizations that have commercial accounts on Facebook, Instagram, and WhatsApp, rather than being limited to content published directly through these platforms.
Legal experts consulted by the site warned that the wording could be used to justify consent for using website content to train artificial intelligence models. Matt Topke, a media and intellectual property lawyer at the Spanish firm "Lowy & Lowy," noted that the clause raises questions about the possibility of accessing high-value commercial materials, the use of which might require paid licenses. However, the experts did not confirm that this interpretation would be accepted by the judiciary.
Following the criticism, Meta removed the controversial wording, confirming that the amendment does not change the mechanism for using automated tools to collect data. This development highlights the legal and ethical challenges associated with using online content to develop artificial intelligence technologies.
The withdrawal of the text does not mean that all issues related to content collection have been resolved; technical access to published materials does not necessarily grant the right to use, reproduce, or incorporate them into model training processes. The legitimacy of these practices depends on applicable laws, the nature of the materials, the terms of access, the rights of their owners, and relevant legal exceptions.
The issue is of particular importance to media organizations that rely on social media platforms to reach audiences, at a time when they are seeking to preserve their rights to content published on their websites. Having a commercial account on a platform may force a publisher to deal with changing terms of use, raising questions about the limits of consent for data collection and the purposes for which it can be used.
These developments come in light of ongoing legal disputes, particularly in the United States, regarding the use of copyrighted materials to train artificial intelligence systems.
In the face of these challenges, there is a need for media organizations to periodically review platform terms of use, regulate crawler access policies to their sites, and clearly define the terms for licensing their content and its use in training artificial intelligence systems.
