Case D-13 · AI Sanctions Case Tracker

Bunce v. Visual Technology Innovations, Inc.

U.S. District Court for the Eastern District of Pennsylvania (attorney Raja Rajan)

Response class R3, Half-Owned It Response class R4, Dodged It Repeat offender (verified in the record) Episode 3

“Any first-semester, first-year law student would know that a fundamental rule of lawyering … is to ensure that any authority to which a lawyer cites does indeed support the proposition for which he cites it.”

The court, in its second sanctions memorandum (Doc. 225, April 20, 2026), in Bunce v. Visual Technology Innovations, Inc.

The case

The tracker's repeat-offender case. Raja Rajan was sanctioned once for AI-hallucinated citations, then filed another motion containing AI-generated citations, in a fight over the costs and sanctions from the first round. The second memorandum is where the court stopped treating the conduct as carelessness: Rajan could not identify any valid reason he had not verified his citations, and the penalty doubled and added compelled education.

What happened

The Eastern District of Pennsylvania's first sanctions memorandum (February 27, 2025) sanctioned Rajan for AI-hallucinated citations in Bunce v. Visual Technology Innovations. He then filed an omnibus motion on costs and sanctions issues that again contained AI-generated citations, which the opposing party flagged.

The rule and the sanction

Rule 11's objective standard, reasonableness under the circumstances; in the Third Circuit, negligence is enough. The first memorandum imposed $2,500 plus one hour of CLE. The second (April 20, 2026) imposed $5,000, plus mandatory CLE courses on AI and legal ethics and proof of the previously ordered CLEs. The court's words: "Any first-semester, first-year law student would know that a fundamental rule of lawyering is to ensure that any authority to which a lawyer cites does indeed support the proposition for which he cites it."

How the lawyer responded

R3, Half-Owned It (the court found the fake cases on its own review; Rajan admitted AI use and the hallucinations in the show-cause response; stated verification is the attorney's responsibility; defended the Eavenson citation; argued sanctions had no deterrent value because he was leaving the case; no remediation recorded; $2,500 penalty and one-hour CLE with no stated credit for his response).

R4, Dodged It (opposing counsel flagged the made-up case; the reply addressed it only in a footnote; the show-cause response argued a reasonable attorney would have filed the motion, blamed opposing counsel's refusal to confer, and argued the Standing Order did not clearly require "human verification only"; he identified no valid reason for non-verification; he asked for a $950 sanction; the court imposed $5,000 and cited his "inability or unwillingness to rectify his behavior"; no remediation recorded). [REPEAT ×2, both episodes in this entry]

How it could have been avoided

A first sanction is a process order, not a warning label. The day it lands, build the verification checklist and run every later filing through it. Re-offending roughly doubles the money and adds the classroom.

Watch the episode

This case is covered in Episode 3: Lawyer Fined Twice for Fake AI Cases in Court (Ep 3) (watch on YouTube).

Sources