Of the 112 bankruptcy, magistrate, district and appellate judges surveyed, 60% said they use AI for their judicial work compared to about 38% who never used any AI tool. Additionally, one in three respondents permit and even encourage the use of AI by those working in their chambers, while 20% formally prohibit the technology, and nearly 18% discourage it.
U.S. Supreme Court justices and U.S. Court of International Trade judges were not surveyed for the study, which will be published in the Sedona Conference Journal.
"The purpose of this study is to understand how, and to what extent, federal judges and other personnel who work in their chambers use AI tools in their judicial work," the report said. "AI is present in federal judicial chambers but not yet a routine, embedded part of most judges' decision-making processes."
The survey, dated March 26, was authored by Anika Jaitley, Daniel W. Linna Jr., V.S. Subrahmanian and Siyu Tao of Northwestern, along with U.S. District Judge Xavier Rodriguez of the Western District of Texas. The findings were published in conjunction with the New York City Bar Association Presidential Task Force on Artificial Intelligence and Digital Technologies.
According to the results, among the judges who use the software, 30% most often use AI for legal research and 15.5% for reviewing documents. Less than 5% use AI to inform or make decisions, transcribe text or audio, or edit and draft documents in cases. Legal staff similarly AI most frequently in legal research and reviewing documents.
Despite the high use, 45% of the surveyed judges said AI training had not been provided by court administration while about 16% said they were unsure any training was provided. Three out of four judges offered training in the software attended those sessions, according to the results.
Judges lean more towards "AI for Law" software rather than general-use models, though the latter is still used heavily. Thomson Reuters' Westlaw AI-Assisted or Deep Research was the most used AI tool, with a reported use of 38%. OpenAI's ChatGPT followed at around 29%, and Thomson Reuters CoCounsel rounded out the top three at nearly 21%.
Funding for legal tech rose sharply last year, driven by AI companies. Total funding increased to $7.08 billion in 2025 from about $4.98 billion in 2024, according to a Law360 Pulse analysis. Notably, no judge reported using the legal AI platform Harvey, which achieved a valuation of $11 billion Friday.
Ultimately, judges are nearly divided between those who are optimistic about AI's potential for the judiciary and those who are concerned, according to the study.
"Many respondents simultaneously recognize AI's potential efficiency gains and express unease about hallucinations, 'zombie cases,' and skill atrophy," the survey said.
The use of AI has come up frequently throughout the federal judiciary, namely for causing hallucinations that have resulted in penalties.
Earlier this month, a Louisiana federal judge sanctioned attorneys for the city of New Orleans over the use of AI that resulted in hallucinated case citations in a pro se civil rights case.
In February, an Ohio federal judge sanctioned two attorneys for repeatedly submitting false and inaccurate citations generated using AI, calling the conduct the most egregious violation of Federal Rule of Civil Procedure 11 he'd seen in his 46 years on the federal bench.
And in January, a federal judge in Pennsylvania reprimanded two attorneys in a copyright infringement suit for filing a motion to dismiss that contained at least eight false case citations generated by AI.
--Editing by Lakshna Mehta.
For a reprint of this article, please contact reprints@law360.com.
