The information contained in legal documents submitted to a court is expected to be accurate. The legal positions advocated are expected to be supported by existing law. This is not something that can be guaranteed when legal documents are generated by artificial intelligence, as has been demonstrated on several recent occasions. 

While there is no general rule regulating how attorneys can use AI to generate legal documents, courts are starting to specifically require documents have human input prior to submission. A federal circuit court of appeals is considering adopting a rule requiring lawyers to disclose the use of AI and certify they have checked the accuracy of any AI-generated material filed with the court. 

Will lawyer have to permanently certify their use of AI in court documents?
Some lawyers have made mistakes with AI – can we prevent these mistakes in the future?

Fifth Circuit Court of Appeals Proposed Amendment to Certificate of Compliance Requires Disclosing Use of AI

That generative AI could not be trusted to produce accurate legal information was first discovered last year when two New York lawyers used ChatGPT for legal research and submitted a legal brief that included fictitious case citations generated by the AI tool. 

Some other courts around the US have responded by banning the use of generative AI or requiring attorneys to disclose whether or not AI was used to draft documents. The highest court to address the issue of using AI to create documents is the Fifth Circuit Court of Appeals. 

In an attempt to ensure legal accuracy and verify human lawyers are the ones practicing law, the Fifth Circuit proposed an amendment to its Rule 32.3 regarding documents submitted to the court. 

The proposed change would require attorneys to certify either that they have used no generative AI to draft any document or, if generative AI was used, that all citations and legal analysis have been reviewed for accuracy and approved by a human. 

A material misrepresentation in the certification of compliance could result in the court striking the document and imposing sanctions on the person signing it. 

The written comment period ended on January 4, 2024, and a final decision on the adoption of the proposed rule is currently pending.

Simply put, AI-generated legal documents may contain false information that the technology just makes up. Why does this happen? It has to do with how AI models learn and process information and their tendency to predict outcomes based on patterns rather than factual accuracy. 

The propensity of generated AI to fabricate information is known as hallucinating. And it’s currently a problem without a solution. 

One of the latest instances of reliance on the unverified accuracy of AI-generated case law citations occurred in December 2023 when the attorney representing (Donald Trump’s former personal attorney) Michael Cohen, submitted phony case citations in a motion to the court. 

Cohen had used Google Bard to do legal research and claimed to be unaware that the case citations he provided to his lawyer could be inaccurate. Cohen’s lawyer apparently accepted the sources provided by his now-disbarred client without bothering to check their validity. 

How Disclosure Requirements Might Affect Attorney Use of AI

The recent incidence of attorneys trusting generative AI to their detriment suggests some lawyers may be using the technology without fully understanding its capabilities and limitations in a law practice. The negative publicity and the court requirements to disclose the use of AI and then vouch for it may deter some law firms from experimenting with content automation in their practices. 

AI is not going anywhere and will continue to develop and become more specialized for various uses – including the practice of law. Lawyers need to become familiar with how AI can best be used to increase the efficiencies of providing legal services while also understanding that AI is not a replacement for professional judgment and must always be reviewed for accuracy. 

The legal community, at least a portion of it, has voiced opposition to the 5th Circuit’s proposal.

Lawyers have expressed concerns, highlighting that existing professional conduct rules sufficiently cover obligations for accuracy in documents. The rise of AI in the legal profession has led to various court orders, some educating about AI, others outright prohibiting it, and most requiring disclosure of AI use.

Bar associations are also examining AI’s ethical implications, with the American Bar Association and state bars like California and Florida actively working on guidelines. The overall sentiment is that AI will significantly transform the legal industry, but there’s caution about explicit rules on AI use in legal practice​​.

A recent article on “Artificial Intelligence (AI) and the Practice of Law” by US District Judge Xavier Rodriguez offers some basic information about AI and how it can be used to benefit a law practice while pointing out some potential issues law firms need to be aware of if they choose to use the technology.

The article acknowledges that AI can be of great assistance to lawyers by performing time-consuming functions such as legal research, document review, and client communications. It may be useful to help with the creation of forms and legal documents. But, any AI output must always be reviewed for accuracy and edited using professional judgment. 

Revealing privileged information is an ethical issue that law firms using AI need to keep in mind. Client confidentiality may be compromised if protected client information is used as a prompt because of how AI may further use the private information. Before submitting client information to an AI platform, a law firm needs to be sure that the information will remain secure. 

As for legal documents filed with a court, Rule 11 of the Federal Rules of Civil Procedure already requires attorneys to certify to the court that a reasonable inquiry was made into the information contained in a submitted document, and based on the inquiry, it is believed the information is accurate and the legal contentions are supported by existing law.

Under the law as it presently exists, it would be rather dubious to argue a reasonable inquiry was conducted when documents that contain fake case citations are submitted to a court. 

The Fifth Circuit’s proposed amendment to its document certification rule doesn’t impose any additional obligations on attorneys who are already expected to present accurate information to a court – including the law cited in support of a legal position. 

Should the Fifth Circuit adopt the proposed amendment and other courts around the country follow suit, lawyers will merely be put on notice that AI is not a stand-in for professional judgment or ethical responsibilities, and attorneys will be held accountable for misuse of the technology.

Written By Mari Gaines, JD – Legal Content Writer

Published by Blue Seven Content

Leave a Reply