The ongoing conflict between artificial intelligence (AI) companies and copyright holders has recently seen significant developments, yet a definitive resolution remains elusive. In mid-February, a U.S. court ruled against Ross Intelligence, a defunct legal research firm, stating it had unlawfully used content from Thomson Reuters’ Westlaw to build a competing platform.
Despite the increasing number of legal cases, the question of whether AI companies can utilize copyrighted content for training their models is far from settled. Carys Craig, an intellectual property professor at York University’s law school, remarked, “We’ve been having this conversation for quite some time already, but it’s still early days.
Generative AI systems, capable of producing text, images, videos, and computer code from simple prompts, require extensive existing content for training. In Canada, a coalition of news publishers, including The Canadian Press, has sued OpenAI, alleging unauthorized use of their content to train ChatGPT. Similarly, major U.S. media companies and the owner of the Toronto Star have filed a copyright infringement lawsuit against Canadian AI company Cohere in a New York court.
Jane Ginsburg, a professor at Columbia University’s law school specializing in intellectual property and technology, noted that only one of approximately 20 to 30 cases has been decided on the merits of the infringement claim and the fair use defense. The remaining cases are in preliminary stages, focusing on procedural issues rather than substantive copyright matters.
While courts interpret existing laws, governments are considering updates to copyright legislation in response to generative AI’s rise. In Canada, creators and publishers are urging the government to address companies using their content for AI training without permission. Conversely, AI companies argue that such use doesn’t violate copyright and that restrictions could hinder AI development in Canada.
The U.K. government is consulting on whether to allow tech firms to use copyrighted material for AI training unless creators explicitly opt out. This proposal has sparked protests from musicians like Elton John and Paul McCartney, with some British newspapers criticizing the government’s approach.
Craig observed that the lack of consensus from consultations means a policy decision will ultimately be necessary, influenced by politics and developments in jurisdictions like the U.S. and Europe. She added that any changes to Canadian law would likely occur after a federal election, potentially leading parties to negotiate licensing agreements amid ongoing uncertainty.
Ginsburg highlighted that more publishers are licensing their content to AI companies, with The Associated Press signing deals with both OpenAI and Google’s Gemini. She suggested that the need for high-quality source data might drive copyright owners and AI companies closer to negotiations.
Source: Swifteradio.com
