Since OpenAI launched its much-anticipated ChatGPT search tool last year, it has spread like wildfire—including in the legal field. The results have been mixed. While some have praised AI for potential efficiency gains in the delivery of legal services, others have raised concerns about the likelihood that relying on AI could result in incorrect or biased content, leading to legal errors.

This issue was brought into the spotlight recently when a U.S. lawyer made headlines for using ChatGPT to find supporting cases for a lawsuit he was pursuing against a Colombian airline. ChatGPT suggested cases and citations that actually did not exist, which the lawyer then used in legal filings, exposing him to the risk of sanctions.

Recognizing this potential for inaccurate and potentially harmful information, some courts and government agencies now require lawyers to disclose if they have used AI-generated materials in preparing their submissions.

ChatGPT and similar AI programs have also taken notice of these risks and started using disclaimers on their sites, warning users of the risks of relying on the results provided by AI. For example, ChatGPT’s homepage warns users that it “may occasionally generate incorrect information.” Similarly, OpenAI’s terms of use includes a disclaimer stating that OpenAI does “not warrant that the services will be uninterrupted, accurate or error free.”

In so doing, providers of AI are putting users on notice that relying on their products, without doing independent verification, could lead to errors or misstatements. The use of disclaimers may be a consequence of user complaints and high-profile news stories like the one discussed above, which reveal the potential risks flowing from this new technology.

What lessons can lawyers learn from AI’s use of disclaimers? Disclaimers can be a particularly effective tool for providers of services to describe what services they are intending—and not intending—to provide. Lawyers can also adopt disclaimers in their own practices to help define the scope of work and limit potential exposure.

What Can Disclaimers Cover?

While the specific risks posed by AI may be relatively new, the use of disclaimers is not. Lawyers have used disclaimers for years in many different ways—for example, to help shape the scope of an attorney-client relationship, to limit the reliance of third-parties on potential advice or to limit their exposure from content posted online or in email communications.

Disclaimers are typically effective when used appropriately. In general, hugely overbroad or vague disclaimers can be less effective than specific, targeted ones—such as disclaimers that expressly state that no attorney-client relationship is intended or that the content (of the communication or post) is not intended to be legal advice.

Imagine a scenario where a lawyer runs a blog opining about issues of real estate law. After a post about a potential legal loophole, a person emails the lawyer to ask a follow-up question about the law. If the person then relies on that blog, or the follow-up email, and the real estate action goes sideways, might they attempt to sue the lawyer for giving them bad advice?

While the lack of a disclaimer doesn’t create liability, the use of one can help limit liability, which is why they are a popular choice. Disclaimers can help reduce or avoid misunderstandings about what a lawyer’s role is and whether an attorney-client relationship exists.

Often, if there is a dispute raised by a person about whether they had an attorney- client relationship with a lawyer, or whether the lawyer was supposed to perform certain tasks, the court will review that issue from the perspective of the (potential) client—was it reasonable for them to believe or expect that they enjoyed the protections of an attorney-client relationship?

A clear disclaimer can help reduce those misunderstandings. In the scenario above, a clear disclaimer on the blog or in the follow-up email—making it clear that no attorney-client relationship is intended or could be created without an engagement letter, or stating that the blog is not meant to be legal advice and that readers should pursue independent legal advice—could help reduce risk or even serve as a bar to the potential client’s relief.

Many states actually require the use of certain types of disclaimers on web pages and social media posts. In Georgia, for example, lawyers are required to present specific and conspicuous disclaimers in communications advertising contingency fee arrangements under the Georgia Rules of Professional Conduct 7.1(a)(5) and (6). Comment [3] to Model Rule of Professional Conduct 7.1 also advises that lawyers can use disclaimers to avoid a finding that a statement is otherwise “likely to create unjustified expectations or otherwise mislead the public.”

Accordingly, while lawyers are generally permitted to publish content like client testimonials and case outcomes, lawyers can limit the potentially misleading effect of such content by including disclaimers that prior results do not guarantee a similar outcome, and that the outcome of any given case depends on the specific facts and circumstances.

Beyond these well-established disclaimers, lawyers can also look to the ChatGPT disclaimer to determine whether there are areas of their own practice that could benefit from the use of similar disclaimers, such as in a legal blog or email signature.

What Does the ChatGPT Disclaimer Mean for Lawyers?

With the disclaimer on prominent display on ChatGPT’s homepage, users of the service are now on notice that the information provided may not be complete or accurate and that ChatGPT is not taking responsibility for verifying the truthfulness of the information.

What’s more, OpenAI’s terms of use include an indemnity provision stating that the user must defend and indemnify OpenAI from any claims arising from the use of the AI services, including the content the user develops in connection with the services.

Regardless of whether such terms are enforceable, they suggest that there can be substantial risk for lawyers in relying on ChatGPT to perform legal services. If a lawyer relies on ChatGPT and submits erroneous legal research to a court, the court (or the client) could find that the lawyer should have known not to rely on ChatGPT for substance in light of the disclaimer and warnings.

Thus, lawyers who use ChatGPT, despite these warnings, might consider reasonable steps to mitigate the risks of their use, such as by independently verifying the generated content, limiting the use of AI to certain low-risk tasks (such as marketing) and staying on top of any obligations they may have to courts, government agencies or others to disclose their use of AI.