){kind=logo}
At Loeb’s AI Summit in Chicago on July 14, 2026, I led two roundtable discussions with in-house lawyers on the trends and challenges companies face when negotiating AI-specific contract terms. We covered a lot of ground, and a few takeaways stood out.
A vendor that can't explain its product is a red flag. Before diving into negotiations, you need to understand what you are actually buying. AI tools vary widely in how they work, what data they touch, whether third-party models are involved and the specific use case. The contract risk profile follows from those answers, but buyers too often find that vendors cannot answer even basic questions about their own products. When that happens, it is often a bigger concern than the contract terms themselves.
What was traditionally considered “market” may be changing. Positions that used to be relatively standard in technology transactions are becoming harder to land when AI is involved. For example, uncapped liability for confidentiality breaches and certain exceptions to consequential damages waivers were once fairly routine asks, but with AI-related risks on everyone’s radar, vendors are pushing back harder. Companies are finding it tougher to hold the line on protections they had historically come to expect.
Signing is the easy part. Organizations invest significant time and effort negotiating AI-specific provisions, but oversight tends to fall off once the contract is signed. There is often a lack of clear processes for tracking whether vendors are actually complying with their commitments, or whether the contract still reflects how the tool is being used. It is a gap many companies acknowledge but few have figured out how to address.
Watch the moving targets. Incorporated online terms that vendors can change unilaterally are a persistent concern with cloud-based solutions, and even more so with AI tools given the complexity of these products and how quickly the underlying technology evolves. A few approaches to manage these risks were discussed, including attaching the applicable terms as of the effective date to lock them in, or adding language that prevents material adverse changes during the term.
Finding leverage in a risk-averse market. AI providers are increasingly cautious about taking on risk, so traditional legal and operational arguments do not always move the needle. Successful negotiations often come down to sources of leverage (e.g., deal size, relationship value and timing) and focusing on what matters most to the business.
The theme across both sessions was clear: know your AI. If you do not understand what you are buying (including the data flows, use rights and third-party tools involved) or how your company plans to use it, it is difficult to effectively negotiate terms that address the risks that actually matter.
