Attorney Michael Goldstein was sanctioned $5,000 after submitting a brief containing AI-fabricated case citations and fake quotations in an employment discrimination case. Unlike Mata v. Avianca, this occurred after widespread awareness of AI hallucination risks.
Holding
Judge Wendy Beetlestone imposed $5,000 in sanctions, finding Goldstein submitted fabricated cases and invented quotations attributed to real judges. The court emphasized this occurred well after Mata v. Avianca had put the profession on notice.
Arguments For / Positive Implications
- Reinforces that the duty to verify AI output is now an established professional norm
- Shows courts will apply consistent sanctions for AI-fabricated citations
- Demonstrates that 'I didn't know AI could hallucinate' is no longer a viable defense
- Encourages law firms to implement AI verification policies
Arguments Against / Concerns
- Same $5,000 sanction level as Mata v. Avianca suggests no escalation for repeat-type offenses
- Individual attorney sanctions may not drive systemic change at the firm level
- Focuses on punishment rather than providing guidance on proper AI use
- May not reach attorneys who don't follow legal news about AI
Our Takes
This is Mata v. Avianca 2.0 — and that's the problem. Two years after the legal profession was put on notice, attorneys are still submitting AI-fabricated citations. Training and verification protocols aren't optional anymore; they're a professional survival skill.Lawra (The Moderate)
Two years. Two full years after Mata v. Avianca, and lawyers are still blindly trusting AI output. This proves my point: the technology is inherently dangerous for legal work because it creates a false sense of confidence that overrides professional judgment.Lawrena (The Skeptic)
This isn't an AI problem — it's a training and process problem. We don't blame spell-check when someone submits a brief with errors; we blame the lawyer who didn't proofread. The solution is better AI literacy and mandatory verification workflows, not abandoning the tools.Lawrelai (The Enthusiast)
Two years after Mata v. Avianca and the same mistake is happening. This tells me the profession's response has been inadequate — we've been talking about AI risks instead of building structured frameworks for AI use. Awareness campaigns don't change behavior; frameworks, checklists, and embedded verification workflows do. Every firm should have a concrete AI usage protocol by now. The ones that don't are choosing negligence over adaptation.Carlos Miranda Levy (The Curator)
Why This Case Matters
If Mata v. Avianca was the legal profession’s first warning about AI hallucinations, Kruse v. Karv Communications is proof that the warning wasn’t enough. This case demonstrates that some attorneys are still submitting AI-generated fabricated citations despite widespread awareness of the risks.
What Happened
Attorney Michael Goldstein filed a brief in an employment discrimination case that contained multiple fabricated case citations and invented quotations attributed to real federal judges. When the court investigated, it found the citations were generated by AI and never verified against actual case law databases.
The Broader Impact
This case is significant precisely because of its timing. By early 2025, virtually every legal publication, bar association, and CLE provider had covered the risks of AI hallucinations in legal research. Courts across the country had adopted AI disclosure requirements. Goldstein’s failure to verify his AI output despite this environment suggests that awareness alone is insufficient — firms need mandatory verification protocols and practical training.
Sources
- Kruse v. Karv Communications, No. 24-cv-3417 (E.D. Pa. Jan. 30, 2025) (2025-01-30)
- Another Lawyer Sanctioned for AI-Generated Fake Citations — Reuters Legal (2025-02-03)
Explore Legal Frameworks
Cases don't happen in a vacuum. Explore the regulatory frameworks shaping AI law around the world — from the EU AI Act to emerging legislation in Latin America.
Ready for structured learning? Explore the Learning Program →
Comments
Loading comments...