AI is becoming integral to the retail industry, with applications in inventory management, demand forecasting, price optimization, content creation and personalized shopping experiences. Companies like Levi Strauss & Co. and Puma are already leveraging AI for customized models and customer styling. As retailers engage with vendors offering AI-powered solutions, either as part of a SaaS offering or as a functional enhancement for existing software, it's crucial to understand and manage the associated risks before diving into negotiations for these agreements.
Understanding AI
AI enables machines to perform tasks that typically require human intelligence, primarily through program structures or layers that are modeled after decision processes in the human brain. There are two key categories of AI: machine learning and deep learning. Machine learning involves structured data and simpler models, which algorithms depend on in order to process data. Meanwhile, deep learning uses complex layers to process large datasets without the need for human intervention. A subset of deep learning is Generative AI (GenAI), which learns from vast amounts of data, including images, audio and other complex data, to generate outputs. Input data is data added to AI to explain a problem, situation, or request. Output data is data generated by the AI based on the input data and AI algorithms.
Risk Management in AI Vendor Agreements
Companies in the retail sector are increasingly using AI to forecast demands for goods, manage supply chains and inventory, and respond to customer inquiries. AI is also being used to cross-sell items during the online buying process based on a customer's shopping history, past purchases, and so on. Equally important, customer trust in the retailers is an important driver in choosing the right AI tools.
When negotiating or reviewing an AI-related vendor agreement, there are numerous issues that retailers may want to consider, particularly with respect to the following standard contract provisions:
- Representations and Warranties
– Does the vendor have the rights to the underlying AI technology, either as the owner or a licensor?
– Will the vendor use your data (inputs/outputs) to train its AI technology? Or in foundational models made available to other vendor customers?
– Does the vendor agree to comply with applicable laws, or at least, agree to support your own AI legal compliance efforts?
– Is the vendor offering audit rights for you to audit its AI training practices? - Intellectual Property
– Will you own (or receive a license in) the AI-generated outputs?
– Are you permitted to use these outputs for internal business purposes only or can they be made available for sale?
– Is the vendor granting you derivative rights in any outputs for purposes of further customization?
– Do you wish to explicitly prohibit the use of your data inputs for vendor AI training? - Privacy
– Will Personally Identifiable Information (PII) of customers become part of your inputs?
– Will any PII be disclosed during, or used in, training the AI?
– What are the vendor's policies with respect to PII?
– Will the vendor agree to robust restrictions on use and disclosure any inputs containing trade secrets, confidential information or PII? - Liability
– Is there a cap on the vendor's liability?
– Are there any instances in which the vendor's liability is uncapped (such as in cases of gross negligence, willful misconduct, breaches of warranties, IP and indemnification obligations)?
– Is vendor liability limited in the event that your output is modified using third-party content in the vendor's AI tool? - Indemnity
– Does the vendor agree to indemnify you for any issues arising from AI outputs, particularly in cases where third-party claims might arise from the datasets being used to train the vendor's AI?
– Does the vendor agree to indemnify you for any issues arising out of the use of the vendor's AI to develop products or services or other uses?
By taking time to understand the implications of these issues, retailers will be better positioned to protect their interests when negotiating with a vendor for AI applications to support their business operations.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
