Navigating AI Risk: Compliance, Contracts, and Emerging Disputes

John Pavolotsky
Partner, Technology & Intellectual Property

Abstract

Artificial intelligence presents new opportunities for businesses, but it also introduces evolving legal and operational risks. In this video, John Pavolotsky discusses key areas where organizations are seeking guidance, including AI compliance, vendor contracting, data ownership, security, and emerging disputes between customers and AI providers. He highlights practical considerations for evaluating AI solutions and managing risk in a rapidly changing regulatory and business environment.

Transcript

I'm John Pavolotsky. I'm a partner in the technology and IP group in the San Francisco office. I'm co-chair of our privacy, cyber, and AI practice.

It is still a very fluid time, and so it is important to track these developments, but it's also worthwhile or useful to understand that it's a rapidly changing landscape, and care needs to be taken to understand what a company's exposure is to potential law.

In terms of AI services provided to clients, we are seeing work in three verticals. The first is compliance, the second is transactional, the third is litigation.

In compliance, clients are asking us, for example, if they're going to be introducing a new chatbot tool, to craft the terms and conditions for that. Comprehending the compliance requirements for that is a necessary part of creating terms and conditions that will ultimately be enforceable and sensible. So that would be one example on the compliance side.

Transactionally, clients, especially with more sensitive workloads, are concerned that vendors do not use that data to train their models, that the data is secure, and additionally, at least from an ownership perspective, that any output generated by these tools is either, as between the customer and vendor, owned by the customer or, inasmuch as ownership of generated output is still a fluid concept, that at the very least the vendor will not claim any ownership interest in the generated output. So those are hot-button areas.

Security is also a hot topic, especially where AI tools are being used, and understanding, comprehending, and ensuring that there's human oversight over agents when they're being deployed.

On the litigation side, we are beginning to see a trend in litigation or disputes between customers and vendors where customers were allegedly promised certain functionality by the AI tool and the promises have rung somewhat hollow.

And so that means, from the customer perspective, being sure that the AI solution is vetted and you understand what you are buying, and from the vendor's perspective, understanding how the solution was developed, what it can and cannot do, and being absolutely crisp, not only in your contract but also in your documentation, as to the functionality of the AI solution.