.
Top Five Unethical Uses of Artificial Intelligence by Attorneys and Four More
August 20, 2026 New York Law Journal
By Jonathan Bick Bick is counsel at Brach Eichler in Roseland, and chairman of the firm’s patent, intellectual property, and information technology group. Adjunct professor at Pace and Rutgers law schools.
As artificial intelligence (AI) becomes more prevalent in legal practice, courts and bar associations increasingly emphasize that attorneys must remain fully responsible for the accuracy, confidentiality, and ethical propriety AI assisted work product. Unfortunately, such intensified emphasis has not resulted in consistent compliance by the legal profession. AI use has interpenetrated every aspect of legal practice, yet the identification and remediation of ethical violations arising from its use have lagged behind.
Auspiciously, the rapid evolution of generative AI has not outpaced the rules of professional responsibility. More specifically, the most common attorney ethical principles implicated related to AI are an attorney’s duty related to competence, confidentiality, candor toward the tribunal, honesty and fairness, supervision of technology and nonlawyer assistance and charging reasonable fees. More specifically, the five most significant ethical risks associated with attorneys’ use of AI follows.
First, certain uses of AI are inherently unethical. One of the clearest and likely the most violated ethical rule occurs when an attorney submits non-public or sensitive client information into a public or open-source AI platform that retains user data for training purposes. Such conduct is a per se, or near-certain, violation of the duty of confidentiality under Model Rule 1.6.
The most frequently cited concern regarding AI in legal practice is its potential to compromise client confidentiality. Model Rule 1.6(a) provides that “[a] lawyer shall not reveal information relating to the representation of a client” unless the client provides informed consent, the disclosure is impliedly authorized to carry out the representation, or another exception applies. Attorneys also have an affirmative duty to exercise reasonable care to prevent confidential information from being inadvertently or unauthorizedly disclosed to third parties. Uploading confidential client information into AI systems that store or reuse user prompts may therefore constitute an unauthorized disclosure.
Second, attorneys act unethically when they rely on AI-generated work product without independent verification. Filing AI-generated legal research, citations, or briefs without confirming their accuracy—particularly when the AI has fabricated authorities through so-called “hallucinations”—violates both the duty of competence under Model Rule 1.1 and the duty of candor toward the tribunal under Model Rule 3.3. The Rules of Professional Responsibility clear state than an attorney is personally responsible for every factual assertion and legal citation submitted to a court, regardless of whether the work product was generated by AI.
Third, attorneys violate their professional responsibilities when they allow AI to make legal judgments or strategic decisions without adequate human supervision. AI is not a substitute for professional legal judgment. Treating AI as an autonomous decision-maker rather than a supervised assistant may violate an attorney’s supervisory responsibilities under Model Rules 5.1 and 5.3. Lawyers must exercise independent professional judgment and ensure that any AI-generated analysis is carefully reviewed before it influences client representation.
Fourth, attorneys may engage in unethical billing practices by charging clients for AI-generated work as though it were entirely attorney-generated. Billing hourly rates for tasks that were substantially completed or dramatically expedited through AI, without making an appropriate adjustment to reflect the actual time and value provided, may violate the requirement that legal fees be reasonable under Model Rule 1.5. While attorneys are entitled to bill for their professional judgment, review, and analysis, they may not misrepresent the amount of attorney labor actually performed.
Fifth, attorneys may fail to obtain informed client consent before using AI in a manner that materially affects the representation. Informed consent is generally required whenever the use of AI involves the disclosure of confidential client information to a third-party platform or when AI plays a significant role in shaping legal advice or litigation strategy. This obligation arises under both Model Rule 1.6 (Confidentiality of Information) and Model Rule 1.4 (Communication).
Absent informed consent, which allow for public disclosure of client information, attorneys must continue to treat all information relating to the representation as confidential, regardless of whether portions of that information may already be publicly available. Clients reasonably expect that confidential information will be shared only within the attorney’s law office or with those reasonably necessary to provide legal services. They generally do not expect that their information will be uploaded into an AI platform for processing.
The elimination or amelioration of these ethical concerns may be achieved by having attorneys discuss their anticipated use of AI with clients at the outset of the representation. This discussion should include the nature of the AI tools being used, the purposes for which they will be employed, whether client information may be transmitted to third-party platforms, and the safeguards implemented to protect confidentiality. Attorneys should also disclose which documents or categories of information may be processed through AI and explain both the benefits and the potential risks associated with AI-assisted legal work.
Failure to disclose how AI will be used, is likely to result in a substantial argument that the client has not provided truly informed consent. Historically attorneys have been duty bound to explain the resources and research methodologies they employ in complex matters so as to promote transparency and hence client informed consent. Obtaining informed consent before incorporating AI into legal representation not only protects client autonomy but also reduces the risk of ethical violations under the Rules of Professional Conduct.
If the lawyer fails to disclose the benefits of AI assisted assistance, it is arguable that the client has not actually demonstrated informed consent. Practitioners may be required to disclose their research capabilities such as access or lack thereof to certain databases, especially in complex cases. It is a good practice to avoid the complications by explaining both the positive and negative aspects of artificial intelligence assisted work product to the client. Following such an explanation, client consent should be obtained and memorialized.
Four less common ethical violations include the following: first, using AI to create or alter evidence. For example, by using generative AI to produce fake emails, contracts, images, audio, or video to support a client’s position. This can amount to fraud, obstruction of justice, or criminal conduct.
Second, using AI improperly to coach witnesses. For example using AI to script testimony or coach witnesses to provide misleading or false statements rather than truthful preparation.
Third, using AI without checking for discrimination may result in an additional ethical violation. Doing so may result in unfairly disadvantage clients or parties based on race, gender, disability, or other protected characteristics.
Fourth, ethical violations may also arise from using AI for deceptive litigation tactics. Using AI to mass-produce frivolous claims, harassing discovery requests, phishing communications, fake online personas, or misleading client communications may violate court rules, ethics rules, and laws against fraud or abuse.