Artificial intelligence is increasingly used by companies to enhance their products, services, processes and decision-making. However, the use of AI also entails legal challenges and risks, especially when it comes to copyright law. This is particularly relevant for M&A transactions, where the buyer needs to assess the value, compliance and potential liabilities or other legal risks of the target company’s AI assets and activities.

One of the main issues is the ownership and licensing of i) the input, ii) the training data and iii) the output of the AI systems. The input and the training data are the materials that are fed into the AI system to enable it to learn and perform certain tasks. The output is the result or the product of the AI system’s operation. All of these elements can be subject to copyright protection, either as original works or as adaptations or derivatives of existing works.

The target company may have obtained the input and training data from various sources, such as public databases, third-party providers, customers, users or employees text and data mining (TDM) or even from internal sources. Depending on the terms and conditions of the source and the way the data is acquired, the target company may have limited or no rights to use, modify, share or transfer the input and the training data for the purposes of the AI system. Moreover, the input and training data may contain personal or sensitive information that may trigger trade secret, data protection or privacy obligations or restrictions.

The output of the AI system may also contain data protected by copyright or other statutory provisions: Output may be similar or identical to the input or the training data, or to other works that the AI system has accessed or processed. It may also be unpredictable, exceeding the expectations or intentions of the target company or the AI system’s developer. In a worst case, the target company may not have the necessary rights or licenses to use, exploit, distribute or transfer the output of the AI system.

A target company’s use of AI may therefore expose it to various liability risks under copyright law, such as claims for infringement, damages, injunctions, disclosure or destruction. These risks could affect not only the target company, but also the buyer, who may assume the target company’s liabilities after completion of the M&A deal. The buyer may also face reputational or regulatory consequences, or lose the expected benefits or synergies of the acquisition. At worst, the target’s failure to comply with applicable IP laws may also “infect” the products or services of the acquirer, if the target’s AI systems (fully or in part) are used for the internal production processes of the acquirer.

To avoid or mitigate these risks, the buyer needs to conduct a thorough due diligence of the target company’s AI assets and activities (“Legal AI Due Diligence”) that covers the following aspects:

• An identification and valuation of the AI systems, their components, their functions and their applications.
• Verification and documentation of the ownership and licensing of the input, the training data and the output of the AI systems, and compliance with the applicable terms and conditions and copyright requirements.
• An assessment and management of the potential liabilities and disputes that might arise from the use of the AI systems, and the availability and adequacy of the insurance, indemnity or warranty provisions or other protections regulated in the acquisition documents.
• Negotiation and drafting of the acquisition documents (in particular: the sale and purchase agreement), ensuring that the buyer obtains the necessary rights and protections regarding the AI assets and activities, and that the seller provides the appropriate representations, warranties and disclosures.

At Dentons, we have the expertise and experience to advise and assist you in all aspects of M&A deals involving AI. We can help you to identify, evaluate and address the legal challenges and risks of AI, and to structure and execute your transaction in a way that maximizes your interests and minimizes your exposure. Contact us today to find out how we can support you in your AI-related M&A deal.

Next, we will take a look at regulatory pitfalls. To be continued!

• Part 2, published on September 16, 2024