5 Key Licensing Considerations for AI Innovations

On Feb. 4, MachiningCloud Inc., a California-based technology company, and its Swiss affiliate, Machining Cloud GmbH, filed a lawsuit against Kennametal Inc. in the Ventura County Superior Court.

The complaint alleges that Kennametal, a multinational industrial tooling corporation, engaged in the misappropriation of trade secrets, breach of contracts and failure to pay contractual commissions exceeding $30 million.

According to the plaintiffs, Kennametal initially engaged MachiningCloud to design and develop a proprietary digital platform — later known as Novo — to support tool selection and purchasing in a rapidly evolving digital marketplace.

The plaintiffs assert that Kennametal unlawfully used confidential information shared during the development of Novo to build a competing platform, subsequently filed a patent application based on that misappropriated technology and withheld commissions tied to over $1.7 billion in sales.

The plaintiffs are seeking more than $330 million in damages, along with injunctive relief to prevent further use or disclosure of the disputed intellectual property.

This case offers a timely backdrop to the broader conversation around licensing innovations as AI-driven innovations may present further challenges.

In the past few years, advancements in artificial intelligence have created many opportunities for businesses. Businesses across almost every industry — ranging from healthcare and finance to manufacturing and media and entertainment — are integrating AI-driven innovations to streamline operations, improve decision-making and maintain competitive advantages.

This evolution has also raised complex intellectual property challenges and questions — particularly in the realm of licensing. Licensing AI-based innovations requires careful consideration of ownership, usage rights and regulatory compliance.

AI-related innovations often involve multiple stakeholders, from data providers and algorithm developers to end users who fine-tune models. As a result, many questions arise: Who owns the rights to an AI-generated output? How should licensing agreements address the situation where the licensee modifies the AI model? What frameworks govern the use of third-party data in AI training?

This article will touch on five key licensing considerations — utilizing hypothetical but realistic scenarios, which businesses will face going forward.

1. Defining Ownership and Authorship of AI-Generated Works

Scenario

A software company develops an AI-powered tool that autonomously generates code. A startup licenses this tool and incorporates its AI-generated outputs into its proprietary software product.

Question

Who owns the AI-generated code — the developer of the AI, the startup using it or neither?

Consideration

Under current legal frameworks, AI itself cannot be recognized as an author or inventor, which means ownership of content in this scenario must be explicitly defined in licensing agreements.

It is important for the agreement to address whether the startup, as the licensee, obtains full rights over the AI-generated code or if the original AI developer retains certain claims.

Without clear licensing terms, disputes over ownership could arise, potentially affecting the startup’s ability to commercialize its product or secure further investment.

2. Managing Data Rights and Compliance

Scenario

A healthcare company licenses its predictive analytics tool to hospitals, promising improved diagnostics and patient outcomes. The AI system relies on patient data to refine its algorithms, but hospitals raise concerns about data privacy, security and compliance with healthcare regulations.

Question

Who controls the data, and how can the hospitals ensure compliance with privacy laws?

Consideration

From an IP and regulatory standpoint, AI licensing agreements must clearly define data rights, usage and compliance responsibilities. Given the sensitive nature of patient data, the agreement should specify how data is collected, stored and shared, ensuring adherence to legal frameworks such as the Health Insurance Portability and Accountability Act.

Another critical consideration is whether AI model improvements derived from a hospital’s data benefit only that hospital or if they enhance the tool for all users. Without clear contractual safeguards, hospitals risk inadvertently contributing proprietary data to a system that enhances others’ capabilities. A well-structured AI license should balance innovation with data security and legal compliance to mitigate these risks.

3. Navigating Open Source Versus Proprietary AI Models

Scenario

A tech company integrates an open-source AI framework into its proprietary chatbot to speed up development and reduce costs. After months of customization and refinement, the company realizes that the open-source license governing the AI framework requires it to disclose any modifications.

Question

Does the company now have to make its proprietary enhancements available to the public?

Consideration

Open-source licensing terms vary widely. This situation highlights the need for businesses to carefully evaluate open-source licenses before incorporating such technology into proprietary systems.

Open-source licenses vary widely in their scope and restrictions, ranging from permissive licenses, such as the MIT or Apache 2.0 licenses, which allow broad commercial use with minimal restrictions, to more restrictive licenses like the GNU general public license, which require that derivative works be made publicly available under the same terms.

Failure to thoroughly assess and comply with the terms of such licenses can expose businesses to unintended consequences, including the risk of the required disclosure of proprietary software, or the inadvertent loss of trade secrets.

To navigate this, businesses should develop a clear open-source strategy that aligns with their IP and commercialization goals. A hybrid licensing model approach may be needed to balance IP protection and licensing needs.

4. Allocating Liability and Indemnification

Scenario

A financial services firm licenses an AI-driven risk assessment tool from a vendor. Relying on the AI’s predictions, the firm makes investment decisions but suffers significant financial losses due to inaccurate outputs.

Question

Is the vendor liable for the erroneous outputs, or is the risk solely on the financial service firm?

Consideration

This situation highlights the role of liability and indemnification clauses in licensing agreements involving AI tools. AI tools can be prone to errors and hallucinations, and licensees and licensors of these tools must determine the extent to which each bears the risk for erroneous outputs — especially in sensitive industries such as the financial and healthcare sectors.

While licensees may seek warranties regarding AI accuracy and compliance, licensors typically limit their liability for AI-generated decisions. The contract should define responsibility for AI errors and establish a fair allocation of risk to avoid costly disputes.

5. AI Patent Licensing and Portfolio Strategies

Scenario

A university research lab patents its AI innovation for drug discovery and licenses it to a pharmaceutical company. The company enhances the algorithm, patents its improvements and integrates them into its proprietary drug development process.

Question

Do these improvements remain tied to the original license, or does the pharmaceutical company now own them outright?

Consideration

Patent licenses involving AI technology must clearly define rights over derivative works and improvements to avoid disputes. In this scenario, the university could utilize a grant-back clause in order to retain rights to enhancements made by the licensee.

On the other hand, licensees may seek exclusive control over their modifications, especially when they involve substantial money, time and resources.

A carefully structured licensing agreement can balance the interests of licensors and licensees by defining ownership rights, use restrictions and commercialization terms.

Conclusion

AI will continue to transform industries and well-structured licensing agreements will be necessary to address ownership, data rights, liability and compliance. Businesses should ensure that their agreements align with evolving laws and industry best practices, while also minimizing risk and maximizing value of their AI innovations.

Lestin Kenton Jr. is a director at Sterne Kessler Goldstein & Fox PLLC.

The opinions expressed are those of the author(s) and do not necessarily reflect the views of their employer, its clients, or Portfolio Media Inc., or any of its or their respective affiliates. This article is for general information purposes and is not intended to be and should not be taken as legal advice.