1. US: Fifth Circuit Sanctions Attorney Over AI Hallucinations
What happened: A Texas attorney was sanctioned by the Fifth Circuit after filing a brief containing fabricated citations and quotations generated by AI. The attorney failed to disclose the use of AI or verify the accuracy of the filing. The "So What": This is a direct signal of malpractice exposure. To protect your firm, you must implement mandatory cite-checking and a "human sign-off" rule for every AI-assisted filing.
2. EU: Don't Pause on AI Act Compliance
What happened: Despite rumors of enforcement delays, EU regulators are urging companies to keep moving forward with AI Act conformity. Guidance from data protection bodies is already shaping how high-risk systems will be classified. The "So What": "Wait-and-see" is now a high-risk strategy. Firms advising EU clients should start inventorying AI systems and risk classifications now to avoid a compliance crunch later this year.
3. Privacy Alert: Claude-Generated Docs Ordered for Production
What happened: A New York court rejected attorney-client privilege claims for 31 documents generated using Claude. The court ruled that even if counsel's information was used in the prompts, the resulting AI outputs were discoverable. The "So What": Prompt hygiene is now a litigation issue. Never put privileged strategy or sensitive client facts into a standard AI prompt—assume those outputs can be forced into discovery.
The Bottom Line: As AI agents become standard in legal workflows, the liability for "agentic errors" remains with the human practitioner.
Is your firm ready? Reply to this email if you need our "90-Day AI Compliance Checklist" for small firms.
The AI Liability Report Research Team