RIGHTSTECH: OPTING IN, OPTING OUT (OF AI TRAINING)

Many creators and rights owners would like to exclude their works from AI training datasets. But simply stating a preference probably is not sufficient given the vast, automated scale at which AI systems operate. Other rights owners are looking to license their archives for training but need to be able to track their use. Startups and content management companies are developing strategies and machine-readable tools to automate the process of opting out, and of opting in. But how effective are they? What are the strengths and limitations of each? Most importantly, will policymakers settle on an opt-in or opt-out standard for the use of copyrighted works in AI, and how should rights owners and developers respond?