This article highlights the significant distinctions to consider when negotiating the use of generative AI software platforms, compared to traditional software solutions. For clarity, we compare (i) generative AI software platforms (whether SaaS-based or installed) with (ii) SaaS and install-based commercial off-the-shelf (COTS) platforms that do not utilize AI or incorporate limited, non-generative AI models.
AI-powered Software-as-a-service (SaaS) and other software platforms have become essential tools for businesses aiming to optimize their operations and services. Basic AI technologies, such as machine learning, natural language processing, and other algorithms (like those used in text editors, autocorrect features, customer service chatbots, search engine optimization tools, and inventory management systems), are now standard in modern software. Consequently, businesses often rely on traditional software licensing principles to negotiate AI software licenses without addressing AI-specific issues. While this approach can still be effective for some platforms, generative AI software (which can create new content like text, images, audio, or other media) necessitates unique considerations.
Tailoring and Personalization
Standard SaaS and COTS Solutions: These platforms are typically standardized products with limited customization options. Unless a customer has specific business needs and sufficient leverage to secure a customized version or a dedicated instance of a SaaS platform, all updates by the provider will apply uniformly across all users. Some SaaS and COTS platforms allow for user configuration within predefined parameters, but significant customization is usually not feasible. Providers also tend to offer standardized services, support, and uptime commitments, with little room for substantial alterations.
Generative AI Solutions: Customization is often a hallmark of many non-generative AI products. Providers of generative AI can work closely with users to adapt models, algorithms, and data processing techniques to meet specific business challenges or to develop unique AI solutions. Therefore, customization becomes a crucial aspect when negotiating AI software contracts, aligning with the user’s specific needs and objectives. This may include defining permissible data inputs, often limiting them to the user’s proprietary data and materials.
Intellectual Property and Data Rights
Traditional SaaS and COTS: IP ownership in these contracts generally pertains to the software itself. The provider retains ownership of the software, granting the user a limited right to use it under specified terms. Users typically retain ownership of their data, though some providers claim rights to user data in aggregated or deidentified forms for service improvement purposes. IP rights related to data generated or processed by these platforms are usually explicitly addressed in the contract.
Generative AI: IP and data ownership issues in generative AI contracts are more complex. Besides ownership of the underlying software, contracts must address ownership of AI-generated content, including copyrightable material like text and images. It may also be necessary to define ownership of algorithms, custom developments, and training data. For instance, if the AI has been trained with proprietary data, ownership must be explicitly outlined. Negotiating appropriate IP and data ownership terms is critical, as these can impact the user’s ability to utilize and develop AI-generated assets. However, IP ownership of AI-generated content in most jurisdictions remains unsettled, meaning that negotiated ownership may only be valid between contracting parties and not enforceable against third parties.
Third-party Intellectual Property Considerations
Standard Software Agreements: These platforms may incorporate third-party IP or technology. The provider is typically responsible for ensuring the platform is non-infringing, legally compliant, and free from harmful code, though the extent of these warranties can vary based on negotiating leverage.
Generative AI Agreements: Generative AI software often uses third-party IP, models, or datasets and generates content based on these materials. Contracts must address third-party IP rights and their impact on usage, including licensing terms, usage rights, warranties, and indemnities. Given the unsettled nature of fair use in AI model training, there is significant risk of infringing third-party IP, particularly if proper limitations on training data are not established. Users may also wish to implement quality controls on data inputs to ensure reliable AI models.
Ethical and Responsible Use of AI
Traditional Software Solutions: Ethical considerations are typically minimal in these contracts due to the limited architecture and data reliance. These platforms usually depend on user-input data rather than third-party data, reducing inherent bias risks.
Generative AI: Ethical and responsible use of AI is a major concern, especially as technology advances. AI use cases and input data can introduce biases reflected in outputs. For example, AI used in HR for hiring decisions must avoid biases in training data to comply with employment laws. AI contracts should establish guidelines for ethical use and regulatory compliance.
Conclusion
Negotiating agreements for generative AI software requires a nuanced approach that goes beyond traditional software licensing principles. The unique capabilities and complexities of generative AI introduce significant considerations around customization, intellectual property, data ownership, third-party IP rights, and ethical use. Businesses must be vigilant in addressing these aspects to ensure their AI solutions align with their specific needs and comply with evolving legal and ethical standards. By carefully negotiating these elements, companies can leverage the transformative potential of generative AI while mitigating risks and protecting their interests.
* * *
This material is for general information only and is not intended to provide legal advice. If you would like to learn more about the topic of this article or our firm’s Technology Law practice, please do not hesitate to contact us at info@danilovpartners.com.