Tech

Seventh Circuit signals sanctions for lawyers using AI to fabricate citations in defamation suit

The US Court of Appeals for the Seventh Circuit has deemed the appeal frivolous, noting that legal submissions contained fake citations and inaccuracies that breached professional standards.

Author
Owen Mercer
Markets and Finance Editor
Published
Draft
Source: Ars Technica · original
Legal fail: Don’t use AI to sue Facebook users for calling you a bad date
Legal team for Nikko D’Ambrosio faces penalties after appeal against Meta and Facebook users relies on fictitious quotes and generative AI errors

The US Court of Appeals for the Seventh Circuit has indicated that sanctions are appropriate against the legal team representing Nikko D’Ambrosio, following a review of an appeal that contained fictitious quotations and fake citations. The court attributed these errors to the misuse of generative artificial intelligence, describing the appeal as frivolous and highlighting judicial frustration with AI-generated legal errors and a lack of professional care.

D’Ambrosio had previously sued Facebook users and Meta for defamation after being criticised in a Chicago dating group. The district court had originally dismissed the case with prejudice, ruling there was no way to amend the complaint. D’Ambrosio appealed anyway, relying on MarcTrent.AI, a firm that claimed to use AI to uncover legal opportunities and increase success rates through predictive modelling.

In a 2025 blog post, MarcTrent founder Marc Trent confirmed the firm had utilised its tech team to draft the initial complaint. He boasted that the firm used everything related to AI and insisted that its technological capabilities would level the playing field against Meta’s legal teams. However, during the appeal, judges agreed that the case was so weak that Section 230 considerations did not even factor in.

Senior circuit judge David Hamilton wrote that the three-judge panel agreed sanctions appeared appropriate. He noted that briefs containing fictitious quotations were unacceptable and bore the hallmarks of generative AI misuse. Hamilton emphasised that inaccuracies discoverable with elementary professional care were not tolerated in court submissions.

The underlying dispute began when Abbigail Rajala, a woman D’Ambrosio briefly dated, posted a screenshot of a text message in a Facebook group. More than two dozen women subsequently shared photos and criticisms of D’Ambrosio. He alleged that the women had doxed him and that Meta profited from the posts, though he provided no evidence of concrete harm or improper contact in the real world.

D’Ambrosio’s legal team attempted to argue that the screenshot was doctored, but the panel rejected this as it was raised extremely late in the litigation. The court also noted that another attorney at the firm, Aaron Walner, failed to sign the filing, which serves as a certification of review. The panel stated that submitting fictitious quotations was inconsistent with the standards of conduct expected from attorneys.

MarcTrent.AI has until June 16 to request a hearing or file statements on whether sanctions are warranted. The firm did not immediately respond to requests for comment. Potential sanctions could include fines covering the costs for the respondents to fight the appeal, as well as penalties for the attorneys involved.

Continue reading

More from Tech

Read next: Netflix secures live broadcast rights for Formula 1 Canadian Grand Prix in US deal with Apple TV
Read next: Silicon Valley’s UBI Push Framed as Technofeudalism in New Critique
Read next: WHO declares Ebola emergency as US imposes travel restrictions and evacuates infected doctor