Public Knowledge Advocates for Fair Use in AI Training, Distinguishing from Infringing Outputs
Public Knowledge, a non-profit organization, is advocating for a nuanced approach to Artificial Intelligence (AI) regulation, particularly concerning copyright law. The organization argues that the training of AI models on copyrighted material should generally be considered fair use, distinct from the acquisition of data or the potential for AI outputs to infringe copyright. Public Knowledge's recent amicus brief in the OpenAI copyright litigation emphasizes that analyzing copyrighted works to learn patterns, similar to how linguists or historians analyze texts, falls under fair use. They contend that a ruling against fair use for AI training would disproportionately benefit large AI and media companies by requiring prohibitive licensing deals, thereby hindering smaller developers and academic researchers. However, Public Knowledge also clarifies that recognizing fair use for training does not excuse companies from responsibility for pirating source material or distributing infringing AI outputs.
