Why are customers suddenly demanding stronger accuracy protections in AI agreements? And why are vendors struggling to provide them?

In this first video, we explore the fundamental shift from traditional software errors to AI-generated hallucinations, and why generative AI creates a completely different type of legal and commercial risk.

For years, technology agreements included standard disclaimers stating that software would not be error-free. Customers generally accepted that reality because software mistakes were usually understood within a familiar framework.
AI changes that dynamic completely.
‍
Generative AI does not simply produce technical bugs or broken functionality.

It generates outputs that seem accurate, appear credible, and often look validated, even when they are completely made up. This poses a more significant risk because users tend to trust the output.

This video explains how hallucinations occur, why they are inherent to probabilistic AI systems, and why this creates tension between customers seeking guarantees and vendors facing technological limitations.

Most importantly, the video introduces the central negotiation challenge that attorneys and contract managers now face in AI agreements:

How do you allocate responsibility for a risk that is serious, commercially important, and impossible to eliminate entirely?

Follow me on LinkedIn or join my newsletter to get updated on the next video in the series, dropping soon. 🚀