Artificial intelligence has been used as a tool for lawyers for years. Lawyers and law firms have successfully (and ethically) used AI programs to help with e-discovery in litigation, for contract analytics, or to analyze statistics from various courts, among other things.

One thing that has evolved more dramatically in recent years, however, is so-called Generative AI (sometimes called GAI). Generative AI is a predictive tool that creates new content—whether as stand-alone material or as a targeted response to an inquiry—based on information it pulls from the internet at large or from specific data pools. Concern about the use of GAI has heightened given recent headlines about lawyers relying on GAI to cite non-existent cases to courts or GAI giving answers that are not correct.

Separate from the issue about whether GAI can assist busy lawyers in their practices, the use of GAI also raises ethical questions. Recently, the ABA Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512, weighing in on what ethical rules are impacted by the use of GAI. Although the ABA recognizes that GAI, as an evolving technology, is a “rapidly moving target,” the Formal Opinion provides guidance for lawyers about the risks to consider.

Here are some tips from the ABA’s Formal Opinion 512.

Rule 1.1—Competence

ABA Model Rule of Professional Conduct 1.1 requires lawyers to provide competent representation to clients, which “requires the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation.” The use of GAI can implicate this rule given the newness of the technology.

Formal Opinion 512 recognizes that lawyers do not have be “experts” in GAI in order to use it in client matters, but “lawyers must have a reasonable understanding of the capabilities and limitations of the specific GAI technology that the lawyer might use.” Like with many other evolving areas of the law, attorneys can demonstrate competence by educating themselves on the technology and its applications, or can rely on the expertise of others who can help guide them. But, the ABA cautions, the technology is fast-evolving, so additional diligence may be required.

In addition, the duty of competence may require lawyers to carefully check GAI outputs before using them for client work. GAI has been known, at times, to generate nonexistent authority, to cite case law inaccurately, or to provide misleading summaries. Because GAI outputs are based on data that is not always reliable or finely-tuned to questions of law, “lawyers’ uncritical reliance on content created by a GAI tool can result in inaccurate legal advice to clients or misleading representations to courts and third parties.” For that reason, lawyers may be obligated to use their judgment and experience to check any materials sourced through GAI. This same risk can also implicate lawyers’ duties to pursue only claims that have some merit and to be candid before tribunals.

For that reason, Formal Opinion 512 recommends that, based on the current state of GAI, lawyers can use GAI as a “springboard” for legal work – but not as a substitute for a lawyer’s judgment and legal acumen.

Rule 1.6—Confidentiality

Model Rule 1.6 protects client confidences from disclosure, calling upon lawyers to keep confidential all “information relating to the representation of a client unless the client gives informed consent.” This rule, in its most common application, prevents lawyers from disclosing confidential or privileged information in a way that puts that protection at risk.

In reviewing whether a lawyer has violated Rule 1.6, courts or bars may look at what information is at issue, whether the lawyer took steps to prevent disclosure, and whether the requirements of confidentiality are too exacting and negatively restrict the lawyer’s ability to provide competent representation to a client.

These issues can become relevant to a lawyer’s use of GAI because GAI tools use – and sometimes retain – the information that has been provided to them. Thus, it is conceivable that a GAI platform that a lawyer employs may not have the appropriate safeguards in place to protect client confidential information from others, given that GAI may “learn” from that information to respond to others’ inquiries.

For that reason, Formal Opinion 512 indicates that before lawyers input information relating to a client in a GAI tool, “they must evaluate the risks that the information will be disclosed to or accessed by others outside the firm.” This may require the lawyer to evaluate how the specific GAI tool they wish to use stores, retains, or uses the information provided to it.

If a lawyer wishes to proceed with disclosing client information to a GAI tool, the lawyer may need to obtain informed client consent. This would require discussion of how the confidential information could become public or known to others; the Formal Opinion provides that “boiler-plate provisions to engagement letters purporting to authorize the lawyer to use GAI [are] not sufficient.” If, on the other hand, the lawyer is using a GAI tool “for idea generation” in a way that does not require the disclosure of confidential information, client informed consent is likely not necessary.

Rule 1.4—Communication

Model Rule 1.4 requires lawyers to keep clients “reasonably informed about the status of the matter.” Although it is extremely fact-dependent, there may be circumstances where a lawyer is required to disclose their use of GAI to clients, and not just when informed consent is required (as discussed above).

The ABA noted that a lawyer should consider whether “knowledge of the lawyer’s use of the GAI tool would affect the client’s evaluation of or confidence in the lawyer’s work.” Although AI tools will continue to evolve, lawyers can help ensure that they comply with their ethical obligations in using AI to assist them in their law practices.