Skip to content

Licensing and AI: Understanding the Challenges of Licensing AI Models

The Schrems Saga Continues: EU Data Protection Authorities May Be Ready to Strike Down Yet Another EU-U.S. Data Transfer Mechanism Background Image

Artificial intelligence (AI) models are transforming industries and the workplace. While few companies will create AI models from scratch, most will find themselves licensing AI models from others. Negotiating those licenses requires a keen understanding of the risks and rewards of AI licensing. AI model licenses begin like any other software agreement, but negotiating those licenses requires a careful eye on AI-specific issues. AI models can be creative, have an enormous need for training data, and vendors often oversell the model’s capabilities. As a result, licensees should consider these challenges during negotiations. Below is a brief introduction to some of these issues.

Intellectual Property

What intellectual property (IP) rights exist in the output of an AI model? Can AI models create trade secrets, copyrights, or even inventive subject matter? If so, who owns the associated IP rights? The owner could be the licensor or licensee (or maybe the model itself?). The answers to these questions are not clearly defined, however, the U.S. Court of Appeals for the Federal Circuit recently confirmed that U.S. patent law requires an “inventor” to be “a human being,” and upheld the U.S. Patent Office’s rejection of patent applications for AI generated inventions. Nevertheless, a licensee can at least define its rights as against the AI model licensor in the license. Licensees should ask that all IP rights created by the AI be assigned to the licensee. For example, the makers of ChatGPT assign all rights to model outputs to their users. If a licensor refuses a similar grant, a licensee should at least ensure that it retains adequate rights to support its business, even after termination. Examples of these rights include, but are not limited to, the rights to reproduce the output, make derivative works of the output, and distribute the output. Otherwise, the licensor could use its rights as bargaining leverage against the licensee.

Performance and Liability

Despite the promise of AI systems, many are still experimental, and AI vendors may over-promise and under-deliver. Licensees should insist on minimum performance metrics to ensure that the licensed model provides adequate accuracy, reliability, and robustness. If the model does not perform as expected, the business consequences may be severe, and could lead to litigation. Licensees should craft warranties and indemnities to ensure that they do not unreasonably bear the risk of underperformance. The licensee should also insist on thorough testing and validation, ideally in advance of the license and on an ongoing basis. Licensees may benefit from a “trial period” to see how the system performs in their business.

Data Protection and Confidentiality

Data is essential to the development of artificial intelligence. As a result, AI vendors often seek to utilize their licensee’s data to improve their products and services. However, many of these vendors lack significant experience or sophistication in privacy and cybersecurity, making data protection and confidentiality even more critical. While data transfers are a standard component of software licenses, using licensee data to improve AI products presents additional privacy and cybersecurity concerns beyond those of a typical software agreement.

Licensees should ensure that any uses of data are consistent with applicable privacy laws and consistent with the privacy notices provided to their users. Numerous states are adopting laws relating to AI and automated decision-making, so licensees will need to stay updated on developments in this area to ensure ongoing compliance.

Further, any transfer of licensee data to the licensor is potentially vulnerable to theft while in the licensor’s hands. Licensees should ensure that licensors take adequate measures to protect licensee data. Some types of data, like protected health information (PHI) and payment card data require specific protections be included in agreements sharing such data. Licensees should consider their own compliance requirements when negotiating the license and ensure the license is consistent with those uses.

What This Means for You

The popularization of AI models brings a complex landscape of legal considerations. As AI becomes commonplace, companies need a solid grasp of these considerations. By staying up to date on AI regulation and seeking the advice of legal counsel, companies can responsibly harness the power of AI technology.

This information is provided by Vinson & Elkins LLP for educational and informational purposes only and is not intended, nor should it be construed, as legal advice.