For lawyers who think they can avoid the conversation, keep in mind that there is an ABA Model Rule of Professional Conduct that binds you here. Model Rule 7.1 prohibits making false or misleading communications regarding a lawyer’s legal services. Letting ChatGPT or a similar product ghostwrite content that a lawyer passes off as their own work product likely falls into this category. The exception might be if a lawyer uses a tool like this for a start but makes substantial changes. The same is true if a lawyer feeds in the lawyer’s template and asks ChatGPT to do some customizations that the lawyer then carefully reviews. But passing off a document wholly created by ChatGPT with virtually no changes is another matter.
But being conservative and erring on the side of informing clients is probably good business and the right thing to do whether it’s ethically required or not. Assuming clients are going to be informed, the next question is how to let them know.
One approach is to be very direct and let clients know that the firm is embracing the technology and then focus on why that decision has been made. Alternatively, a firm could choose to gradually introduce the topic by using the software for smaller, less consequential tasks and letting the client get used to the software. And then when the time is right and the client understands how the technology works, roll it out for more significant tasks.
With respect to the “all in” approach, clients may appreciate a firm’s honesty, but they’re going to want answers to several questions (and those using the gradual approach will still likely get these questions, even if they’re asked over time). In-house counsel will want to know how a firm using the software is complying with legal ethics rules.
1. What is ChatGPT? Why does it matter?
Lawyers shouldn’t assume their clients are all informed about the technology (unless, of course, you’re solely representing clients in high tech). Provide easy to digest materials explaining how the firm can use the technology and what it can and cannot do. Maybe offer video tutorials and provide examples of how you’ve used the technology for other clients or what other firms are doing.
2. How are attorneys and paralegals using the software being trained (Rules 5.1 and 5.3) and how is the AI itself being supervised?
Firms should have protocols in place to ensure lawyers and staff understand the software’s functionality and its weaknesses. And work produced by the AI needs to be carefully reviewed. Stories in the media about “hallucinations” where the software makes up facts and then states those misstatements in a very convincing way are not mere anecdotes. Generative AI tools are getting better at addressing this, but no one has solved the problem entirely.
3. How are you ensuring my client confidentiality and data privacy are being secured?
The law firm should highlight data security measures being taken to protect the client. One thing firms can do is show that they are not putting any confidential data into the software or, alternatively, the firm is using third-party software connecting with OpenAI (or a similar product) using an application programming interface (API) key. For example, popular ChatGPT Outlook and Word plugin GhostWrite founder, Patrick Husting, explains:
There are several advantages to having your own OpenAI API key. Firstly, it allows for customization and fine-tuning of AI models to better suit your specific needs. Additionally, having your own API key ensures that your usage is isolated from other users, thus minimizing the risk of unauthorized access to your sensitive information. Furthermore, acquiring your own API key enables you to monitor and track usage, which is particularly important for businesses that need to comply with data protection regulations.
4. What benefits can I expect to see as a result of using the technology?
There are several benefits that may appeal to clients. First, the software can potentially help speed up your work on client matters. If you can produce a document in a day instead of a week, that’s a major benefit for the client. You can potentially argue that in some ways it can do a better job than an attorney. If you’re using a legal research product that relies on generative AI (like Casetext’s CoCounsel), talk about how the software can potentially do a better job of finding answers than a human can without this kind of technology.
And, of course, you should talk about how the cost savings of using the software will help the client. If you bill on an hourly basis, you might be able to pass some or all the savings on to the client (where on that spectrum will depend on where you strike the balance in your interests between being less expensive or more profitable). For firms billing on a flat-fee basis, you might be able to lower your flat fees or offer your clients a lot more accessibility since you’ll have freed up time.
It might be worth noting as well, particularly for litigators, that opposing counsel is probably using the technology and you are interested in being evenly matched when it comes to the technology being used to represent your client.
5. Do you have to use it for my matters?
Firms should offer clients an opt-out option allowing them to direct the firm not to use the technology in their matters.
These approaches are more about the timing of your communications and not about whether you implement using generative AI or not. Whichever approach is taken, there are several issues to consider when communicating with clients about your firm using generative AI.
A successful lawyer is one who has clients that trust in him or her and having this conversation is important. Clients may assume the technology is being used and be upset that they have not heard from you. And when they do hear from you, a thoughtful response addressing their concerns will reinforce your image as being forthright and innovative.