Generative AI, a groundbreaking technology, holds the potential to revolutionize content creation, accelerate development, and streamline tasks for companies. However, as with any powerful tool, it comes with both opportunities and risks. In this article, we delve into the crucial contractual aspects that require consideration when dealing with providers of generative AI tools. The focus is on safeguarding your company’s confidential information, intellectual property, and overall business interests within the contract terms.
Defining generative AI
Generative AI refers to artificial intelligence technology trained to discern patterns in training data and apply that knowledge to generate new content based on user prompts. This content spans various formats, including text, software code, data, images, video, and audio.
Exploring the potential and risks
Generative AI offers a plethora of applications and benefits, but it’s not without risks. To better understand both sides of the coin, consult our “Top Ten Considerations for Companies Using Generative AI Internally.”
The importance of agreements
Your use of a generative AI tool hinges on an agreement with the tool provider. These agreements typically manifest as online terms of service or tailored contracts. They define the rights and obligations of each party, including ownership and usage rights for prompts and outputs, confidentiality clauses, and legal terms such as indemnities and liability limitations. For simplicity, we’ll refer to these agreements as “terms of use.”
Differentiation in agreements
Certain providers extend “enterprise” or “business” versions of their tools with more favorable terms of use than standard agreements. Moreover, the negotiation room might exist to craft a specialized agreement more suited to your company’s needs. While these avenues might entail extra fees, they could provide crucial safeguards compared to the standard, one-size-fits-all approach.
Varied service categories
Generative AI providers often offer distinct service categories, such as plugins or APIs that integrate with the “base” tool. These integrations might be subject to differing terms of use. They enable seamless technology integration, but they might also necessitate closer examination of the associated terms.
Understanding rights in prompts
While terms of use often stipulate that end users own their prompts, providers might retain extensive usage rights, including creating derivative works and sharing prompts with other users. In some cases, unless users opt for a premium account, providers can utilize prompts to refine their AI models. This has implications for sensitive company information, potentially indirectly benefiting competitors. Establishing a generative AI use policy can help manage these risks.
Navigating rights in output
Similar to prompts, end users generally own generated outputs as per terms of use. However, providers could retain rights to distribute or employ these outputs for training their AI models. If relinquishing these rights poses issues for your company, exploring private enterprise accounts, opt-out mechanisms, or refining your use cases might be necessary.
Many terms of use exempt providers from confidentiality obligations regarding user prompts and outputs. This could leave your company vulnerable if confidential information is disclosed. Sharing third-party confidential information might breach nondisclosure agreements, and including personal data in prompts could violate privacy laws. Furthermore, these actions might compromise your stance in trade secret disputes. Crafting a comprehensive generative AI use policy can raise awareness about these risks among your team.
Indemnification considerations
Generative AI tool providers typically don’t indemnify end users. However, some enterprise or business accounts might offer indemnification against claims stemming from third-party intellectual property infringement. Conversely, many standard terms of use saddle end users with broad indemnification responsibilities. This could lead to substantial obligations for your company while affording limited recourse against the provider.
Liability exclusions and limitations in generative AI tools’ terms of use often shield providers from substantial contractual liability. These may encompass disclaimers for indirect, incidental, and consequential damages, with caps on direct damages. Enhanced liability caps might be available for enterprise or business accounts. In contrast, end users usually have limited exclusions and liability limitations. Unless customized terms are negotiated, your company could face significant liability risks with little protection from the provider.
Generative AI tools hold exciting prospects for businesses, but comprehending both the potential advantages and legal pitfalls is pivotal. Evaluating associated technical, business, and legal risks will guide your decision-making process. Conducting a thorough assessment of potential generative AI tools, their risks, and the contractual terms provided by their vendors is essential to harnessing this transformative technology safely and effectively.
