OpenAI Urges U.S. Govt. to Codify ‘Fair Use’ in AI Training Amid Legal Battles Over Copyrighted Content

OpenAI has urged the U.S. government to formally codify ‘fair use’ as a legal standard for AI training, reinforcing its position that AI models should be allowed to learn from copyrighted material without explicit permission.

In a policy submission for the “AI Action Plan”—an initiative under the Trump administration aimed at reshaping U.S. AI policy—OpenAI argued that the doctrine of fair use has been instrumental in America’s dominance in AI research and innovation.

“America has so many AI startups, attracts so much investment, and has made so many research breakthroughs largely because the fair use doctrine promotes AI development,” OpenAI stated in its submission.

However, the proposal comes at a time when OpenAI is embroiled in legal battles over its own use of copyrighted content, with some of the largest media organizations and creative professionals accusing the AI firm of illegally training its models on protected works.

The Legal Battle Between OpenAI and The New York Times

One of the most high-profile copyright lawsuits against OpenAI was filed by The New York Times (NYT) in December 2023, marking a pivotal moment in the ongoing debate over AI training and intellectual property rights.

In its lawsuit, The New York Times accused OpenAI of “mass copyright infringement”, alleging that its articles were used without permission to train AI models like ChatGPT. The lawsuit provided multiple instances where OpenAI’s chatbot was able to regurgitate near-verbatim passages from Times articles when prompted. Some outputs even included editorial content that was behind the newspaper’s paywall, raising concerns that OpenAI’s data scraping practices had bypassed content restrictions.

According to The Times, OpenAI’s use of its content directly competes with the newspaper by providing summarized news and analysis without driving traffic to its website. The lawsuit seeks monetary damages and a court ruling that forces OpenAI to stop using copyrighted content without authorization.

In response, OpenAI defended itself by arguing that its model does not store articles but rather “learns from large datasets in a transformative way,” a common argument in fair use cases. The AI firm has also hinted at reaching licensing agreements with publishers to settle copyright disputes, though it has yet to reach a deal with The New York Times.

The lawsuit has significant implications for AI companies, as a ruling against OpenAI could set a precedent requiring AI developers to pay for content usage rights, potentially increasing the cost of AI model training and limiting access to vast amounts of data.

OpenAI’s Double Standard on Copyright: The DeepSeek Controversy

Interestingly, OpenAI’s own stance on copyright infringement has been inconsistent, as seen in its reaction to DeepSeek, an AI model developed by a Chinese research group.

When DeepSeek was released in early 2025, OpenAI expressed concerns that the model had used its own proprietary data without permission. This raised eyebrows in the AI and legal communities, as OpenAI has repeatedly defended its own use of third-party copyrighted material, claiming it falls under fair use.

The hypocrisy in OpenAI’s stance did not go unnoticed. Critics pointed out that if OpenAI expects protection for its own proprietary work, then the same principle should apply to media organizations and creators whose content is used in training AI models.

This incident further fueled debates over whether AI companies should be allowed to freely scrape the internet for training data or whether stricter licensing and copyright protections should be enforced.

How Other Countries Are Handling AI and Copyright

The U.S. is not the only jurisdiction grappling with the AI copyright dilemma. Governments worldwide are racing to establish legal frameworks that balance AI innovation with intellectual property protection. For instance:

• European Union: The EU has taken a strict approach, with the AI Act requiring AI companies to disclose what copyrighted materials were used in training datasets. This level of transparency is something U.S. lawmakers have yet to enforce.
• Japan: Japan has adopted a more flexible stance, allowing AI developers to train models on copyrighted material as long as the output does not directly compete with the original work.
• China: Chinese authorities have mandated that AI companies must obtain explicit permission before using copyrighted content, a stark contrast to OpenAI’s reliance on broad fair use arguments.

OpenAI’s call for a codified fair use doctrine will likely face strong opposition from media organizations, creative professionals, and publishers who argue that AI companies are profiting off their work without compensation.

With multiple lawsuits pending and policymakers still debating AI copyright laws, the outcome of this battle will shape the future of AI development, content creation, and intellectual property rights.

If OpenAI succeeds in its push for expanded fair use, it could pave the way for unrestricted AI training, benefiting AI developers but potentially undermining the financial viability of the media, publishing, and creative industries. However, if courts side with content creators, AI firms could be forced to license copyrighted materials, changing the way AI models are trained and potentially limiting their capabilities.