AI “Hallucinations” Are No Longer Just Your Problem—Courts Are Targeting the Other Side Too

Courts are no longer satisfied with just sanctioning the lawyer who files AI-fabricated authority. The emerging trend is more pointed—and more dangerous for litigators: judges are increasingly criticizing, and penalizing, opposing counsel who fail to catch it.

Recent decisions reflect growing frustration on the part of the judiciary with fake citations generated by AI. While the “offending” attorney still bears primary responsibility, courts are now withholding fee awards, denying relief, and issuing admonitions where opposing counsel did not identify and raise the defect. The takeaway is blunt: failing to spot obvious errors is becoming a problem for all concerned.

Competence Now Includes Policing the Record

Courts are moving toward a view that the adversarial system requires “shared vigilance.” Judges have made clear they should not be the last line of defense against fictional authority and have openly questioned why counsel did not perform even a basic cite-check of opposing filings. Appellate courts have echoed that sentiment, noting that failures on both sides to catch inaccuracies undermine the process and waste judicial resources.

What was once discretionary diligence—spot-checking only suspicious citations—is quickly becoming an expected baseline. If the other side cites it, you are now expected to verify it.

The Seventh Circuit’s recent decision in Dec v. Mullin is illustrative. Therein, the court admonished petitioner’s counsel after her brief included two non-existent case citations and a false quotation bearing the “tell-tale signs” of AI hallucinations. Although the court declined further sanctions because counsel accepted responsibility and the errors appeared limited, it also noted that opposing counsel’s failure to catch the fabricated authority “gives us pause.” The message is unmistakable: courts expect all lawyers—not just the lawyer who filed the defective brief—to be vigilant about hallucinated authority.

The California Court of Appeal reached a similar conclusion in Noland v. Land of the Free, L.P., where plaintiff’s counsel filed appellate briefs filled with fabricated AI-generated quotations and citations. The court imposed a $10,000 sanction payable to the court and referred the matter to the State Bar, but it declined to award attorney fee sanctions to opposing counsel because they had not alerted the court to the hallucinated authorities and appeared to recognize the problem only after the court raised it. In other words, even where the filing attorney’s misconduct was sanctionable, the court was unwilling to reward opposing counsel for missing the same problem.

The same concern surfaced again during appellate argument in Landberg v. City of New York  (oral argument video at minute 20). When the court confronted counsel about hallucinated or otherwise non-existent authority, the panel did not stop with the lawyer who cited it. Presiding Justice Hector LaSalle also pressed opposing counsel for failing to alert the court, explaining that the challenged cases were used as part of the argument and that it was “striking” opposing counsel did not bring the problem to the panel’s attention. The exchange underscores the point: once AI-generated citation errors enter the record, courts may expect both sides to help clean them up.

Professional Liability Exposure Is Expanding

From a professional liability perspective, this is where the risk sharpens. The theory is no longer limited to “you filed it, you own it.” Courts are beginning to signal that you ignored it, you may own that too. Failure to identify fabricated authority can now be framed as lack of competence, lack of diligence, or a failure to meaningfully respond to the opposing party’s arguments. Even without formal sanctions, courts have already shown a willingness to deny otherwise valid relief based on that lapse alone.  The practical consequence is a second-layer exposure: your case can be right, but your process can still cost your client.

For Ohio lawyers, nothing about this trend requires a new rule. The existing framework is already sufficient—and already being emphasized. Under the Ohio Rules of Professional Conduct, AI is expressly treated as “relevant technology,” folding it into the duty of competence, Rule 1.1. That means understanding and accounting for AI hallucinations. See also Prof.Cond.R. 3.3, candor to the tribunal, and Rules 5.1 and 5.3, supervisory obligations.

Bottom Line

Courts are recalibrating expectations in the AI era. The lawyer who files hallucinated authority remains the primary target. However, the lawyer who fails to catch it can no longer claim unclean hands.