Jim Milliot , Publishers Weekly; AI Challenges, Freedom to Read Top AAP Annual Meeting Discussions
"The search for methods of reining in technology companies’ unauthorized copying of copyrighted materials to build generative AI models was the primary theme of this year's annual meeting of the Association of American Publishers, held May 9 over Zoom...
“To protect society, we will need a forward-thinking scheme of legal rules and enforcement authority across numerous jurisdictions and disciplines—not only intellectual property, but also national security, trade, privacy, consumer protection, and human rights, to name a few,” Pallante said. “And we will need ethical conduct.”...
Newton-Rex began in the generative AI space in 2010, and now leads the Fairly Trained, which launched in January as a nonprofit that seeks to certify AI companies that don't train models on copyrighted work without creators’ consent (Pallante is an advisor for the company.) He founded the nonprofit after leaving a tech company, Stability, that declined to use a licensing model to get permission to use copyrighted materials in training. Stability, Newton-Rex said, “argues that you can train on whatever you want. And it's a fair use in the United States, and I think this is not only incorrect, but I think it's ethically unforgivable. And I think we have to fight it with everything we have.”
“The old rules of copyright are gone,” said Maria Ressa, cofounder of the online news company Rappler and winner of the 2021 Nobel Peace Prize, in her keynote. “We are literally standing on the rubble of the world that was. If we don’t recognize it, we can’t rebuild it.”
Ressa added that, in a social media world drowning in misinformation and manipulation, “it is crucial that we get back to facts.” Messa advised publishers to “hold the line” in protecting their IP, and to continue to defend the importance of truth: “You cannot have rule of law if you do not have integrity of facts.”"