Mar 14, 2025 10:43:00

OpenAI says AI race ends if training on copyrighted work isn't fair use

OpenAI

has been sued by various companies, including The New York Times, for using various content on the Internet to train its AI. However, OpenAI has stated that using copyrighted works to train its AI 'would be the end of the AI race if it were not fair use .'

OpenAI declares AI race “over” if training on copyrighted works isn't fair use - Ars Technica
https://arstechnica.com/tech-policy/2025/03/openai-urges-trump-either-settle-ai-copyright-debate-or-lose-ai-race-to-china/

OpenAI argues that President Donald Trump's AI Action Plan, scheduled to be released in July 2025, will resolve the debate over whether using copyrighted content to train AI is fair use and is essential for the United States to win the AI race against China.

At the time of writing, a court case had been ongoing regarding whether using copyrighted content to train AI was fair use. Content creators argued that using content to train AI was a copyright infringement, as it would reduce the value of the entire creative work. On the other hand, OpenAI argued that even if copyrighted content was used to train AI, the output of the AI would be different from the original content.

So far, AI-related lawsuits have been decided in favor of copyright holders, with courts ruling that using copyrighted content to train AI is not fair use because legal research firm Westlaw found that AI-created content could replace original content in the marketplace.

In response, OpenAI is pinning its hopes on President Trump's AI Action Plan to avoid a similar ruling. 'OpenAI's AI models are trained not to replicate content for public consumption. Instead, they learn from content and extract patterns, language structure, and contextual insights,' OpenAI said. 'That means the training of our AI models is consistent with the core objectives of copyright and fair use principles, using existing works to create something entirely new and different without diminishing the commercial value of the existing work,' OpenAI argued, arguing that its use of data constitutes fair use.

On March 13, 2025, OpenAI said that President Trump's AI Action Plan requires the United States to shift its copyright strategy to promote 'freedom of learning' in the AI industry, ending the legal battle over whether the use of copyrighted content for training AI is fair use. Furthermore, OpenAI argued that if the legal battle is not ended, Chinese companies will continue to have access to copyrighted content that American companies cannot access, giving China an advantage.

In addition, OpenAI said, 'By preserving the ability of American AI models to train on copyrighted material, the federal government can ensure Americans' freedom to learn from AI and avoid losing their AI advantage over China.'

'If Chinese developers have free access to data while American companies do not have fair use access, the AI race will effectively be over,' he said. 'If the United States loses the AI race to China, it will also mark the end of the success of democratic AI. Ultimately, access to more data from the widest possible sources will ensure access to more powerful innovations that yield more knowledge.'