The Wild West of Law: AI ignites ethics debate in the legal profession

Peering over the landscape of law, generative artificial intelligence emerges on the horizon. Ask any lawyer their thoughts on the topic and most have a quick answer, though the consensus differs. Is it a useful tool worth exploring or a mirage putting both attorneys and clients at risk? Some are already integrating AI while others bide their time until the dust settles. Whatever AI’s potential might be, it’s the Wild West out here.

“In a few years, I suspect everyone will be using it in some form or fashion,” said attorney Austin Fax of Lowther Johnson Attorneys at Law LLC. “It’s just a question of when, not if.”

The American Bar Association established its Task Force on Law and Artificial Intelligence in 2023 and issued ethics guidance on AI use the following year. The ABA Model Rules of Professional Conduct require lawyers to “keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology.” Although Missouri has yet to issue formal guidelines, the Office of Legal Ethics Counsel and Advisory Committee of the Supreme Court of Missouri issued an informal opinion in 2024. It suggests lawyers should follow “duty of competence” when deciding how and when to use AI, “maintain requisite knowledge and skill” of the technology, understand risks and benefits, consider confidentiality and not rely solely on AI information.

For those legal professionals who do integrate AI, use and platforms vary. Aaron Wynn, an attorney with Brad Bradshaw MD JD of Kansas City LLC, said his firm has integrated AI for both legal and nonlegal tasks. He uses general AI platforms for tasks such as making a business trip itinerary but relies on CoCounsel AI by Thomson Reuters for legal tasks such as summarizing documents or creating case event timelines.

Patrick Douglas of Douglas, Haun, & Heidemann PC researched several legal-specific AI programs before subscribing to ChatGPT for small tasks such as providing a starting point for research, creating initial drafts, catching grammar mistakes and identifying potential concerns in documents.

“It is very helpful in terms of just routine things,” he said. “You could upload an 80-page PDF document with text searchability and, in that scenario, it could break down very complex documents to hone in on things you want to get to quickly.”

While AI can check attorneys’ work, attorneys must always check AI’s work, Douglas said: “If people use it blindingly and don’t check the underlying citations of where is this info coming from, that is where you get bitten. … You are ultimately responsible for what you’re going to tell a court or another party or, for that matter, a client. You own it, so you better make sure it’s right.”

Jason Primuth is the chief innovation officer of Lexitas, which utilizes AI technology to provide legal transcription for firms nationwide, including in Springfield. He also offers education on AI ethics for the legal profession. While some argue AI bots could remove a human touch from the legal process, Primuth wagers it actually removes robotic tasks to allow more room for attorneys to do the things only they can.

“At a high level, AI is really good at digesting text,” he said. “It takes a human several hours to do that kind of thing and, for us, we can do it in a matter of minutes.”

Wynn uses AI in such a manner: “The purpose of AI should be to do a better and more efficient job for the people you’re trying to help. … In my job, it has enabled me to do that, with focusing more on my clients than ancillary matters.”

Lexitas’ technology is specifically for legal use and licensed through OpenAI. His team has offered these services since April 2024 with six months of beta testing assessing the outputs of 50,000 transcripts from more than 100,000 attorneys – all of which he said came back accurate. When OpenAI’s GPT-4 took the uniform bar exam in 2023 as a test, it passed in the 90th percentile. Vice president of CoCounsel at Thomson Reuters, Pablo Arredondo, said in a Q&A with Stanford Law School that GPT-4 appears to be the first AI large language model “smart enough” to be utilized in professional settings. In the civil case Muhammad Iqbal v. Zayad in Pakistan, GPT-4 was found by the International Bar Association to have provided results consistent with Pakistani law.

“We’re helping attorneys do their jobs and helping clients have better trust in their attorneys. Humans miss stuff. And I think what we’re providing helps humans, attorneys, paralegals do a better job with their cases,” Primuth said. “It can improve confidence that the work attorneys have done has been double- and triple-checked.”

Although CoCounsel AI was created specifically for the legal profession, Wynn still scrubs documents of client information before uploading to ensure confidential information isn’t shared and subsequently stored in the platform.

“There are some lines I’ve drawn for myself,” he said, adding that AI is simply a tool to help get toward a desired end. “It’s me thinking it through and the bot helping me get there. … You are the professional. You have to review this. You have to cross reference and check. This is your work; you just want to make it faster.”

New York lawyers Steven A. Schwartz and Peter LoDuca were hit with sanctions for creating Federal District Court documents with hallucinated or fabricated content from ChatGPT – including fake case law. Similarly, layperson Roberto Mata sued the airline Avianca for injury during a 2019 flight. His lawyers provided a 10-page brief that cited ChatGPT-hallucinated case law.

However, the greatest ethical concern with AI is the ethics of client confidentiality. For Syd Tippie, assistant attorney general for the Missouri Attorney General’s Office of Springfield and member of the Board of Governors of the Missouri Bar, the ethical risk does not outweigh the benefits at this time. He does not use AI, noting that his thoughts are not necessarily representative of his employer or the Missouri Bar.

“I think it really lands them in a lot of liability,” he said. “It’s not a position I’d want to be in, potentially putting my licensure on the table for convenience.”

Tippe can see potential benefits of AI in the future, with tasks such as proofing briefs. However, because the burgeoning technology does not have direct Missouri guidelines at this time, the unknowns are too great – especially when it comes to how data is stored or compounded to create potentially biased outputs.

However, there is always a risk when using third-party vendors, Douglas said: “You’re making judgments all the time about who is trustworthy in that regard. So much of what you’re doing already involves a component of trust of your third-party vendor and their process.”

Fax agrees with Tippe. Although he is open to using AI once it’s more developed, he has chosen to opt out at this time.

“At some point, we’re probably going to see something that is maybe a great consensus in the legal community of what sort of software platforms are and are not good in terms of both ethics and functionality. Right now, it’s a bit like the Wild West,” he said. “I think ultimately it will end up being another tool in the attorney’s toolbox.”

In a September 2024 address at the Missouri Bar’s annual meeting, Hon. Mary R. Russell, chief justice of the Missouri Supreme Court discussed AI and encouraged attorneys to become educated on the topic: “The bottom line is this: Generative AI is here to stay, and its use in our profession is inevitable. We all have practical and ethical duties to become competent about generative AI. But we also have an imperative to establish appropriate guardrails around its use in our legal profession.”

Cases like the New York lawyers are the exception to the rule, Douglas suggested, saying that most attorneys will take the time to review AI outputs like they would any other documents. Tippie said he hopes to reframe these cases by focusing on the courts that issued sanctions to attorneys who misused AI.

“What we’re seeing in that situation is the integrity of the court,” he said. “The court is overseeing these things and handling it in the most appropriate way it can.”

Wynn said he hopes to see further AI guidelines issued by the state, enabling more attorneys to safely utilize it as a tool: “If people want to feel safe and secure industrywide, we need the Missouri Bar to say how to use these systems. Until then, there is going to be an inherent level of risk that would be mitigated by an industrywide adoption for those things.”