Liability Considerations In Enterprise Use of Generative AI

ABOUT THE PROGRAM

Since ChatGPT burst onto the scene last fall, developers of large language and other foundation models have raced to release new versions; the number of app developers building on top of the models has mushroomed; and companies large and small have considered—and reconsidered—approaches to integrating generative AI tools within their businesses. With these decisions has come a cascade of practical business risks, and copyright and copyright-adjacent issues have taken center stage. After all, if your marketing team’s Midjourney-like AI image generator outputs artwork later accused of infringement, who is ultimately responsible? And how can you mitigate that risk—through contractual indemnity? through guardrails deployed in your training process? through post-hoc content moderation?

To closely examine these questions from all perspectives, our panel brings together lawyers and technologists with experience at AI model creators, at AI deployers, and at a global web platform considering how best to navigate the rapid changes wrought by generative AI. To these voices, we add one from a computer science researcher and legal scholar who has investigated conditions under which the current generation of large language models runs the risk of reproducing copyrighted material.

Our program has two major parts. In the first half, we focus on commercial realities—on what in-house counsel have seen on the ground; on questions they have fielded and advice they have given about potential uses of ChatGPT, Midjourney, and other generative AI tools; on contractual indemnities they have requested and received.

In the second half, we turn to a discussion of legal liability doctrines and examine the extent to which current law protects AI deployers and users and the extent to which adaptation may be needed to accommodate generative AI. We will examine classic copyright-centric doctrines, like secondary liability and the DMCA’s section 512 safe harbor as applied to content moderation. But we will also zoom out to other online intermediary liability issues many enterprises will no doubt face, such as claims for defamation or the provision of false or inaccurate information resulting from generated AI output, and the extent to which the Communication Decency Act’s Section 230 safe harbor can—or perhaps in some modified form should—offer liability protection.

It is a jam-packed 90 minutes, but you will come away with a thorough grounding in the core copyright and adjacent liability considerations for deploying generative AI tools in your enterprise.