A new study suggests artificial intelligence may already be influencing the written opinions of federal appellate courts across the United States.
Attorney Josh Morrow ran approximately 2,250 published federal appellate opinions from 2026 through an AI-detection tool, with more than 50 returning signs of AI-generated writing.
To establish a baseline, Morrow also tested more than 300 published circuit opinions from January 2022, and the detection tool flagged none of those documents.
The contrast between the two datasets is striking and points toward a meaningful shift in how at least some judicial chambers may be approaching the drafting process.
Morrow has been careful to stop short of claiming his results conclusively establish that AI was used, but they do offer substantive evidence worth taking seriously.
The study may also capture only a fraction of actual AI involvement, since the technology can be used to brainstorm, research, outline, critique drafts, or test counterarguments without leaving obvious traces in the final opinion.
Those more subtle applications would be difficult or impossible for any detection tool to identify, meaning the real extent of AI use in chambers could be considerably broader than the numbers suggest.
For the legal profession, the findings raise a set of pressing practical questions that practitioners and scholars are only beginning to grapple with in any structured way.
Lawyers must now consider how to read and evaluate opinions that may have been shaped, at least in part, by an AI system rather than purely by a judge or their clerks.
Deeper questions follow around whether AI involvement changes how the profession thinks about legal writing, the weight of precedent, research methodology, and the kinds of arguments attorneys choose to present before courts.
The possibility that judicial opinions could carry the fingerprints of machine-generated language challenges long-held assumptions about authorship and accountability in the legal system.
As AI tools become faster, cheaper, and more capable, pressure on courts to manage growing caseloads may make these technologies increasingly attractive, regardless of formal guidance or policy.
The legal community faces an urgent need to establish clear standards around disclosure, verification, and the appropriate boundaries of AI assistance within the judiciary.