Credit: Depositphotos
Background The consumer rights advocacy company DoNotPay has decided to halt efforts to put their “robot lawyer” in court. On January 20th, CEO and founder of DoNotPay, Joshua Browder, announced on Twitter that his company would be using their artificial intelligence (AI) software to instruct an attorney and their client (the defendant) through the contest of a traffic ticket in court. The attorney and defendant were planning to wear earpieces in the courtroom that would allow the AI to listen to the court proceedings and instruct them with real-time responses, acting as what would have been the first AI-led defense in American courts. Just five days after announcing their plans to implement their “robot” lawyer in court, Browder reported that he would be canceling the stunt, claiming that the State Bar prosecutors had threatened him with six months of jail time if he went through with his plan.
Since the company’s founding in 2015, DoNotPay has offered an expanding array of services focused on consumer protection, ranging from assistance canceling free trials to filing Freedom of Information Act (FOIA) requests. The company utilizes AI chatbots to assist consumers in filling out forms and corresponding with organizations when trying to assert consumer rights. DoNotPay’s goal is to provide more accessible legal services to consumers who may not have the financial resources to hire private attorneys to assist them in defending their rights against large corporations. The company boasts an eccentric array of investors, from the Chief Product Officer at Adobe, Scott Belsky, to the musician group, The Chainsmokers.
DoNotPay’s AI lawyer is a new and more sophisticated AI than the company had used before. The company uses OpenAI's GPT-3 API which is a later version of ChatGPT produced by the same company. It appears that DoNotPay is combining OpenAI’s GPT-3 with their own learning models to produce a sophisticated chatbot with knowledge of the legal field.
As AI has grown more powerful and popular, the subject of AI in the legal field is being discussed more often, receiving mixed responses. Some are embracing the usefulness of AI in law, while others raise concerns about letting a chatbot practice in a field with an important emphasis on ethical standards.
Legal Implications
This traffic court stunt was not Browder’s first attempt to use an AI robot lawyer in an actual courtroom. Earlier this year, Browder offered a $1,000,000 reward to any attorney willing to wear an AirPod and repeat the AI-derived responses in the U.S. Supreme Court. His Twitter feed was quickly filled with replies from attorneys providing a myriad of reasons why his plan would fail. Browder somewhat jocularly replied, “Ah yes how could I forget about the AirPods Security Checkpoint." Unfortunately for DoNotPay’s founder and CEO, the Supreme Court has a strict ban on all electronic devices in the courtroom. Until the highest court in the country decides to amend this rule, it remains impossible to use AI beyond preparations outside of the courtroom.
Browder was initially hesitant about relegating his grand plan to a lower court due to concerns that his “haters” would criticize traffic court as being too easy for ChatGPT. However, with his Supreme Court plans on hold until its electronic rules are changed, his team used AI in municipal traffic court. But, again, they were presented with legal hurdles. Multiple undisclosed state bar officials reminded Browder that the unauthorized practice of law is a misdemeanor offense punishable by up to six months in jail.
The State Bar of California defines the unauthorized practice of law (UPL) as the act of offering legal services by someone who is not licensed to practice law, which is a criminal offense. This law aims to ensure that clients seeking legal representation hire competent attorneys who understand the law. Leah Wilson, Executive Director of the California State Bar, told NPR that the recent spike in AI-empowered legal services raises concerns about how they should be regulated. State Bar officials and courts must use their existing powers to limit the use of AI-driven litigation tools until new regulations are passed.
In addition, the use of litigation-centric AI presents challenges to attorney-client privilege. Recently, Amazon’s general counsel warned its employees to cease inputting confidential information about the company into ChatGPT following observations of it mimicking Amazon’s internal data. Although DoNotPay relies on an API to utilize ChatGPT, more conflicts of interest could arise when both parties in a dispute use AI tools without understanding the implications of inputting sensitive information to the service. This threatens attorney-client privilege and could jeopardize the client’s ability to openly and honestly discuss their legal matters.
Strict professional and ethical standards also bind attorneys. However, AI continues to demonstrate implicit biases that could affect the fair administration of legal services. AI models such as ChatGPT perpetuate and amplify existing inequalities when trained on discriminatory data-sets, in addition to human input and computational biases. If Browder aims to democratize legal services to the 80% of low-income individuals who cannot afford legal representation, many of whom are minorities, then addressing the implicit biases of AI is crucial to best serve clients, while also complying with the professional rules of conduct.
Business Implications
Although their current numbers are not public, DoNotPay was founded in 2015 and quickly reached a valuation of $210 million by 2021. Following their recent setbacks, Browder is refocusing the company’s efforts away from litigation back towards transactional work, helping consumers address high medical bills, unwanted subscriptions, and credit reporting agencies. The public nature of DoNotPay’s recent stunts may also spark regulators to take action against AI legal tech. Despite this, Browder remains optimistic that “history will be made.” While this may be true, it is unlikely to occur in the immediate future.
*The views expressed in this article do not represent the views of Santa Clara University.