A federal judge in February ruled that a criminal defendant's chat with the AI Claude was not protected by attorney-client privilege. That decision has sharpened concerns inside law firms that wealthy clients who use ChatGPT and other chatbots could waive privilege. Partners at major firms say clients are using AI to research strategies, upload trust documents and even challenge legal advice. Lawyers warn the practice can increase billable hours, erode trust and expose sensitive files.

Clients bring AI into legal planning

High-net-worth clients are asking their lawyers about recommendations they got from AI. Tasha Dickinson, a partner at Day Pitney, said she fields calls every week. Some clients admit using chatbots. Others conceal it, she said, but their questions give it away.

Dickinson described one case involving a Florida client. The client told her AI suggested a community property trust to save taxes for heirs. Dickinson realized the client had a different situation. His wife had recently died. A community property trust is a tool for married couples. It was not suitable for him. "You do understand that a community property trust is between husband and wife, right?" she asked. The caller fell silent.

Lawyers say those conversations are more than embarrassment. They change how work gets done. Robert Strauss, a partner at Weinstock Manion, said clients have uploaded trust documents into AI systems and returned with lists of suggested edits. Strauss said the firm ends up spending extra hours responding to those AI-driven questions. "We end up spending two, three, four hours of time dealing with stuff that so far has amounted to nothing," he said. "I haven't actually received a single workable suggestion from that process."

Privacy and privilege concerns

Beyond wasted time, lawyers warn about legal risk. When clients feed documents into large-language models, they may be waiving protections. Dickinson said her main worry is that sharing privileged documents with an AI can forfeit attorney-client privilege.

She said that risk keeps her awake at night.

Those concerns now have a legal precedent. In February, a federal judge held that a criminal defendant's chat with Claude about his defense strategy was not protected by attorney-client privilege. The ruling is narrow, but it signals how courts may treat communications routed through commercial AI services. Lawyers who represent wealthy clients say the ruling changed the tone of conversations inside firms.

Mastheads

Privacy is another front. Strauss said the act of uploading sensitive estate planning papers, tax records or medical information into third-party AI systems creates data exposure. He warned that those systems may store or process inputs in ways clients don’t expect. That makes people wonder about confidentiality, regulatory duties and the handling of client data under professional rules.

How law firms are responding

Firms are adjusting policies. Strauss said his firm is revising client engagement agreements to include explicit warnings about AI use. The draft language will tell clients that entering privileged materials into public or commercial chatbots may void privilege. It will also clarify whether the firm will accept documents that have been through a model or whether the firm will require fresh originals.

Some lawyers are taking a different tack. They use AI themselves as an internal research tool. They run documents through approved, secure models or use software with contractual guarantees about data handling. Firms that adopt those tools are trying to balance efficiency with control over confidential inputs.

Still, lawyers who work with wealthy clients say managing client expectations is becoming routine. They explain the limits of AI. They flag how a generic suggestion from a chatbot may not apply to a client’s specific facts. And they warn about the paperwork and privilege issues that can follow.

Workload and client relationships

AI is changing the shape of legal work even when it doesn't generate a usable idea. Strauss described a familiar pattern: clients read an AI output, return with a set of proposed edits or alternative clauses, and ask the firm to justify its drafting choices. The firm then must re-explain its legal reasoning. That takes time. It also shifts focus from legal problem solving to managing misunderstandings.

Dickinson said AI can be useful in some cases. It helps clients learn basic concepts and frame questions. But she said it becomes a headache when it replaces careful client interviews and document review. Cases that require nuance and a detailed factual record don't map cleanly onto the shortcuts AI offers.

The surge in client-side AI use touches rules of professional conduct. Lawyers have duties to protect client confidences. They also have obligations to supervise nonlawyer assistants and to avoid conflicts of interest. Feeding privileged documents into an external model raises all three issues, lawyers say.

Regulators haven't issued a uniform rule. But the February court decision shows judges are willing to consider whether a communication retained privilege after passing through an AI vendor. That case involved a criminal defendant and Claude. It didn't create a broad ban on AI. It did, however, provide a concrete example lawyers now cite when advising clients.

Law firms are increasingly trying to steer clients toward safer practices. Several firms are drafting language that explains the trade-offs of AI research. Firms want clients to understand that convenience comes with consequences.

Practices include asking clients to certify whether a document has been uploaded to an AI tool. Firms may require clients to provide unaltered originals or to acknowledge that their use of a public chatbot could waive privilege. Some lawyers plan to refuse work if key files have already been shared with a commercial model.

Related Articles

"I haven't actually received a single workable suggestion from that process," said Robert Strauss, partner at Weinstock Manion.

This article was created with AI assistance.