The sample size and methodology of the Bloomberg Law report have not been verified, and this publication has not reviewed the full report. The Above the Law column is a short data roundup built on it, and its companion commentary describes a Texas federal judge who criticized lawyers over suspected AI-drafted briefs and a Kentucky judge who weighed AI's role more carefully. The survey is not isolated. Legal Cheek reported in February that only one in five lawyers placed high trust in AI-generated work, and Morae reported in July that only 33% of senior legal professionals trusted AI-assisted results. A Supio survey of 207 U.S. personal-injury practitioners found that 99% would not use AI content they could not verify.
For practitioners, the takeaway is verification. Every citation and quotation needs to be checked before filing, and firms should track how courts respond to AI-drafted briefs, since judicial scrutiny is already shaping what counts as acceptable use. The gap between heavy daily use and low confidence means the real cost of AI may sit in review time, not in drafting speed, so supervising partners should measure that time before assuming AI is saving hours.