AI and Intellectual Property: Legal Issues in a New Era for Your Business

Artificial intelligence (AI) is rapidly transforming the intellectual world, but its impact extends beyond just functionality.

The legal landscape surrounding intellectual property (IP) challenges brought on by AI’s unique capabilities will be resolved in the courts overtime, as the hysteria of yet another new technology causes risk, opportunity and ambiguity. Some issues that impact your business and what you can do about should be known by every organization.

Authorship and Inventorship: Who Gets the Credit (and Rights)?

Current IP law hinges on the concept of a human author or inventor. Copyright law, for instance, protects the “original works of authorship” created by humans in Title 17 of the United States Code. However, AI can now generate creative outputs like music, text, and art. This raises the question: can AI itself be an author?

The US Copyright Office and courts currently maintains that AI cannot hold copyright. This may not apply in the future with artificial general intelligence (AGI) is a type of AI that matches or surpasses human capabilities across a wide range of cognitive tasks. Current AI is narrow AI which is designed for specific tasks like financial forecasting or drafting text. This leaves the question of who owns the rights to AI-generated works.  In some cases, the programmer or the entity that trained the AI might be considered the author. However, recent cases like Stephen Thaler v. US Patent and Trademark Office challenge this notion, blurring the lines of inventorship.

AI and Copyright Infringement: Training on Borrowed Ideas

AI systems use what is called a “training token,” the more tokens, theoretically, the better the output, perhaps the more “original it is.” Large language model AIs (like Gemini, or Chat-GPT) are often trained on massive datasets that may include copyrighted material. This raises concerns about potential copyright infringement.  For instance, an AI trained on a corpus of copyrighted music might inadvertently reproduce elements of those works in its own creations. Legal frameworks are still being developed to address whether using copyrighted material for AI training falls under fair use or constitutes infringement. The plaintiffs like The New York Times, maintain anything an AI creates from training data is a “derivative work”, a work based on another work of authorship,  that is one of the six exclusive rights copyright holders enjoy.

The Ownership Conundrum: Who Controls AI-Generated IP?

Even if the authorship issue is resolved, ownership of AI-generated IP remains unclear.  Companies that develop AI tools might argue for ownership, while users who interact with the tools and provide prompts might also have claims.  Additionally, the question of whether the underlying data used to train the AI should factor into ownership rights adds another layer of complexity. Authors have been claiming ownership of AI text and images for years.

The Path Forward for You and Your Business

IP law needs to adapt to the realities of AI for your business. Potential solutions include:

• Ensuring your terms of use are updated to remove any ambiguity in the law. This creates a contractual right rather than relying on statutory or common law.
• Ensure you know the AI’s training token architecture if you rely on AI for your business. More and more deployments of AI allow the organization to point to the training data for the AI.
• Do not rely on a “fair use” defense without consulting and attorney as this is a multifactor text and can be complex and sometimes, subjective.

As AI technology continues to advance, ongoing legal discussions will be crucial in fostering innovation while ensuring fairness and protecting the rights of creators and developers. For more assistance on how to craft an AI policy or contracts that stand the test of time, contact Buckley Law attorney Martin Medeiros at 503-620-8900.