Confidentiality + AI

Can the parties put confidential information into AI tools? Increasingly, parties are addressing this in their confidentiality and protective orders.

Here's a recent example:

4.2 Use of Generative AI. No party or authorized recipient may input, upload, or submit Confidential Material into any artificial intelligence application, including any generative, analytical, or large language model-based tool (“AI”), unless the application is prohibited from storing or using the input Confidential Material to train or improve its model and the Confidential Material is able to be and is removed and deleted consistent with Section 10. A party intending to use AI that it contends meets these requirements must retain written documentation of these contractual protections.

I like this one.

It puts the parties on notice of what I believe is essentially an ethical obligation to understand how the chosen tool(s) train or improve its models based on your (or the other side's) content. By requiring these conversations with IT or the tools' builders themselves, we gain a better understanding of what we're dealing with and ensure some protections against disclosure to third parties.

It also aligns with where I think we'll land when all is said and done: using AI tools as an extension of an attorney, in the process of litigating and for the purposes of case development, is allowed, doesn't waive privilege, and is covered by the attorney work product doctrine--so long as certain protections are in place.

🔥✌🏻♥️