chevron-down Created with Sketch Beta.

ARTICLE

Ten Things You Need to Do Before Using Generative AI

Michael P Daly and Meaghan Geatens

Ten Things You Need to Do Before Using Generative AI
Nitat Termmee via Getty Images

Like computer-assisted legal research and document review, generative artificial intelligence (AI) is the latest in a long line of tools with the potential to make lawyers more efficient and more effective. But there is a fundamental difference between using traditional AI to find something and using generative AI to draft something; whereas the former might miss something that does exist, the latter might cite something that doesn’t.

As recent headlines have made clear, the latter can at minimum cause considerable reputational harm.

For that and other reasons, using generative AI implicates several ethical duties. Some—like competence and candor—have received considerable attention. Others—like confidentiality and communication—are less apparent but no less important. Fortunately, the ABA recently issued Formal Opinion 512, which offers guidance for those considering whether to use generative AI. From that and other opinions like it, several best practices are emerging.

Check Your AI Program’s Instructions, Terms of Use, and Privacy Policies

Understanding a program’s capabilities (and limitations) is crucial to competently using it (and to obtaining a client’s informed consent to your using it). Be sure to have at least a general understanding of how it works, how it stores and shares the information you input, and how reliable its output is.

Check Your Client’s Engagement Letter

Check your engagement letter (and any incorporated policies and procedures) to see if a client has agreed to the use of generative AI. For consent to be informed, the client should be told how an AI tool might use the client’s information, and how you might use the AI tool’s output.

Check Your Firm’s Policies & Procedures

Some firms have been quick to embrace generative AI, even going so far as to build their own AI tools. Others have prohibited using generative AI for client work. And many have yet to announce AI policies, but may still have security policies that may be implicated—even for personal work on personal devices. Be sure to check your firm’s policies before using generative AI.

Check Your Court’s Local Rules and Judge’s Orders

Some courts and judges have updated their local rules and standing orders to address the use of AI tools. Whereas some judges have prohibited its use (which begs the question of whether courts are empowered to ban the use of a tool), others have allowed its use but required lawyers to disclose their use or attest to having checked the accuracy of their filings. Check the court’s local rules and judge’s standing orders to verify what is restricted and required.

Check Your Local Bar Association’s Guidance

Much like courts, local bar associations have also issued guidance and best practices for using tools. When educating yourself about how tools work and what’s permissible, check to see if your local bar association has issued any ethics opinions about the use of generative AI in your jurisdiction.

Check Your Bills

If you plan to charge for generative AI services, make sure to explain that in advance. And, as always, lawyers who bill on an hourly basis must bill only for the actual time spent (whether that is for inputting information into a tool, or reviewing it afterward), and should ensure that their charges are reasonable.

Check Your Input

Avoid uploading any information (for example client information, personally identifiable information, or privileged communications) that should be protected from disclosure. Even if a program is secured within your firm, your colleagues may not know that its output was originally confidential. Confirm that your uploads don’t implicate any data-privacy concerns.

Check Your Output

Do not prompt an AI tool to generate content that could infringe on a third party’s intellectual-property rights. That could include trademarks or likenesses, among other things.

Check Your Citations

At a minimum, the Model Rules require that lawyers not only know that generative AI can hallucinate, but also check to make sure it hasn’t. Before relying on work produced by generative AI, verify that its output is accurate; ultimately, lawyers are responsible for their work regardless of how it was produced.

Check Your Instincts

Finally, use your better judgment. If it is used at all, generative AI should be used to assist lawyers, not replace them. Consider whether its use is appropriate for your task in your circumstances.

    Authors