Companies implementing AI tools internally and those that want to sell AI products face an uncertain regulatory future. State legislators have begun working on their versions of AI regulation, with 25 states proposing artificial intelligence bills in 2023.

In-house counsel should consider these three steps to put their companies in a position to traverse the uncertain regulatory landscape:

1. Documentation: Robust documentation practices that record information about every AI model and every relevant dataset are critical. Documentation will allow legal teams to understand how each version of your AI tool existed at a particular time. Documentation should include logging each version change of data sets and AI models as they develop over time, as well as appropriate corporate policies for internal AI use.
2. Pick the right tool for the job: Generative AI has countless use cases, but there may be better tools for your AI need. Sometimes, simple AI models trained on a specific task may suffice. It is essential to talk with your internal technical team to pick the right tool for the job without exposing your company to potential liability based on the usage of AI models that may be subject to higher regulatory scrutiny.
3. User agreements: Having user agreements with AI tools is crucial for informing users about the software's operations and the data it utilizes, as well as mitigating the risk of illegal use. As proposed regulations increasingly focus on user rights and protections, it is essential to keep users informed about your AI capabilities and to protect against any misuse by the user.