[ad_1]
The body of questions revolving around copyright law and AI is, with no doubt, one of the pressing IP issues of interest as we dive headlong into 2024. Even the United States Copyright Office recently concluded a request for public comments as part of “a study of the copyright law and policy issues raised by artificial intelligence systems.” Are you curious about how many comments the Copyright Office received? Over 10,000 comments were sent in, with all of them available for your perusal and consideration at your leisure. In particular, I commend the comments submitted by some of legal academia’s most august copyright experts, Pamela Samuelson, Christopher Jon Sprigman, and Matthew Sag, whose comprehensive responses to the Copyright Office’s Notice of Inquiry serve as a very useful primer on the breadth and depth of issues implicated by the water-and-oil interplay between copyright and AI. Readers would also benefit from the response by Samuelson, Sprigman, and Sag to the FTC’s comments, including their critique of the FTC’s attempt to usurp the “judicial function” of deciding “what is and what is not copyright infringement, or what is or is not fair use” with its assumption that training of AI systems using copyrighted information would represent actionable copyright infringement.” In the view of the authors, that remains an open question that will require further judicial input over the coming years.
Besides the interest by legal academics in the matter, there was a range of industry responses to the Notice of Inquiry as well, including submissions from ChatGPT purveyor OpenAI on one end, and major publishers like the New York Times on the other. The latter, as with other publishers, expressed concerns about copyright violations by AI providers, with respect to at least two scenarios. First, publishers have expressed concerns about using copyrighted materials to “train” AI systems, making the argument that such unauthorized training is a copyright violation. Or to put it differently, that companies like OpenAI should be paying publishers for the right to train their AI systems with copyrighted materials. Second, publishers have expressed concerns about the output of AI systems, arguing that there are clear examples of copyright infringement where an AI system responds to a user prompt with an answer that is substantially similar to the copyrighted material that the AI system was trained on.
For its part, OpenAI has expressed at least some sympathy for the concerns raised by publishers — up to a point. As the company reports, it has entered into “early partnerships with the Associated Press, Axel Springer, American Journalism Project and NYU” that it believes shows its intent to act as a good citizen vis-à-vis the news industry. In the Axel Springer deal, for example, OpenAI negotiated “access to its news archive” that “allowed the artificial intelligence firm to use newly published articles in apps like ChatGPT. The deal, which includes a “performance fee” based on how much OpenAI uses its content, is worth more than $10 million per year” according to reports. Additionally, even though OpenAI also maintains that using copyrighted materials to train services like ChatGPT is fair use under copyright law, it still provides publishers an “opt-out” that stops OpenAI’s “tools from accessing their sites.”
Despite OpenAI’s legal positioning and steps to act responsibly when it comes to copyright issues in this new and uncertain legal and technological landscape, it has proven unable to avoid being hauled into court to defend itself based on copyright claims advanced by disgruntled publishers. In the most recent and prominent example, OpenAI was sued by the New York Times, apparently after talks between the companies about a license arrangement broke down — probably because the parties couldn’t agree on a fair price for unfettered access to the Times’ content. Sued alongside OpenAI was Microsoft, which incorporates AI-sourced material in its Bing Chat search engine product. As expected, the Times argues that OpenAI’s training of ChatGPT is not a fair use but copyright infringement, and that ChatGPT and Bing Chat will return infringing responses to user prompts based on the Times’ copyrighted material. While OpenAI has until the end of February to respond to the Times’ complaint based on a waiver of service, its responses have already received coverage in the tech media.
Ultimately, there is a long way to go until we see how courts respond to the various and sundry copyright lawsuits filed against OpenAI, including the case just filed by the Times. But, no doubt, these cases will continue to receive massive amounts of both popular and IP media attention, due in no small part to the potential of AI to revolutionize our lives. Whether or not we end up with a situation where OpenAI continues to strike one-off deals with publishers to use their content — while simultaneously arguing that it doesn’t really need to do so to avoid copyright infringement exposure — these cases promise to forge new ground in our understanding of the limits of copyright law in the face of a rapidly changing technological landscape. As these cases work their way through the U.S. legal system, feel free to ask ChatGPT for a prediction as to how they will turn out.
Please feel free to send comments or questions to me at gkroub@kskiplaw.com or via Twitter: @gkroub. Any topic suggestions or thoughts are most welcome.
Gaston Kroub lives in Brooklyn and is a founding partner of Kroub, Silbersher & Kolmykov PLLC, an intellectual property litigation boutique, and Markman Advisors LLC, a leading consultancy on patent issues for the investment community. Gaston’s practice focuses on intellectual property litigation and related counseling, with a strong focus on patent matters. You can reach him at gkroub@kskiplaw.com or follow him on Twitter: @gkroub.
[ad_2]
