AI professionals are closely monitoring how AI companies should manage the content used for training their models since the launch of AI Overviews. Google has recently disclosed its approach, emphasizing fair utilization, offering opt-out choices, and underscoring paid arrangements for certain circumstances.
Google’s recent policy paper argues that using publicly accessible web data to train models should be safeguarded under fair use in the U.S. The company suggests opt-out controls and current copyright regulations as ways to address publisher worries.
The document “A Practical Strategy for AI Regulation in the United States” compiles the viewpoints previously presented by Google. It addresses the current push from regulators and publishers for greater control, including clearer acknowledgment and, in some cases, payment. This resource provides valuable guidance for publishers navigating AI usage in content distribution by outlining Google’s stance on the matter.
Google’s stance on copyright
Google compares AI training to an art student gaining inspiration from viewing artwork in a gallery. It also proposes that the same degree of protection should be applied globally with text-and-data-mining exceptions.
Google suggests that website owners employ machine-readable controls such as Google-Extended in their robots.txt if they wish to prevent their content from being utilized. When AI replicates existing content, the key is not to assess whether the output is “too similar,” but to follow established notice-and-takedown procedures as detailed in the paper.
Google is exploring innovative methods to generate value by collaborating with websites that offer content to ensure AI responses are current and precise, as well as striking deals to acquire access to exclusive content. The article does not mention any specific initiatives, conditions, or timeframes.
Where the Position Ends
The UK’s CMA has implemented a new rule allowing websites to choose not to use AI search features and mandating Google to credit publisher content. This move aims to enhance publishers’ negotiating leverage. Google has begun testing an opt-out switch, but publisher reports currently lack click data to assist in decision-making.
US publishers are reinforcing their position by sending a cease and desist letter to the Common Crawl Foundation. They emphasize that seeking permission is necessary before using content, opposing the opt-out approach mentioned in Google’s paper.
Why This is Important
Google is pushing to maintain its current strategy amid discussions about new regulations.
Publishers and regulators are requesting additional features beyond the current paper offerings, such as compensation, permission-based scraping, and detailed click-level data. In turn, the paper is responding by providing controls and managing negotiations on a case-by-case basis.
Looking to the future
These are policy stances, not promises of specific products. The partnerships and content agreements discussed by Google could impact how publishers benefit, but the specifics remain open to interpretation. Monitor if Google connects its programs, conditions, or numbers with the value-exchange language found in its current policy papers.
Featured Image: FotoField/Shutterstock
