Does an AI receptionist put attorney-client
privilege at risk?
It can if handled carelessly. That is why firm direction, retention, access controls, and attorney review matter.
Vanessa is deployed as your firm's agent, the same role a receptionist or intake paralegal has always filled. Below is how that structure is intended to support confidentiality, what the cited cases say, and where the analysis remains unsettled.
Privilege protects the conversation, never the facts.
Attorney-client privilege protects the communication between a client and the firm. It never protects the underlying facts: the crash, the date, the injury are always discoverable, no matter who first wrote them down. The goal is not to suggest that intake creates a new shield. It is to minimize what exists and protect the channel it travels through.
That is the whole design philosophy here: keep the minimum, label it clearly, and route it only inside the firm.
How firm direction and attorney review support confidentiality.
Vanessa is deployed as the firm's agent, the same role a receptionist, intake paralegal, interpreter, or answering service has traditionally filled. That structure is intended to support privilege, but privilege is fact-specific and ultimately determined by a court.
A lawyer's agent falls within the attorney-client privilege where the agent's role is to help the lawyer serve the client.
Confidentiality duties attach to prospective clients, even callers who never retain the firm.
What the courts actually said.
The cited 2026 rulings involve different facts and do not establish a universal rule for AI intake. They distinguish consumer use from tools directed by counsel, but privilege remains fact-specific.
In the cited decision, the court distinguished consumer use from a tool directed by counsel and discussed how an AI system might be analyzed as counsel's agent. The case did not involve Vetted Intake.
"AI platforms are tools, not persons." Using one did not waive work-product protection. The decision did not involve Vetted Intake, but it supports analyzing AI as a tool rather than a person.
A consumer AI allegedly gave legal advice and told a user to fire her lawyer. The lesson we built in from day one: intake only, no legal advice, a human attorney always in the loop.
Layered protection, not a promise.
Assume any recording, transcript, or AI summary could one day be discovered. So we minimize retention, label every AI output "unverified, attorney to confirm," and configure and validate per-matter legal-hold controls during a signed implementation when the firm requires them. Because this area is developing, we present these safeguards as layered protection, not a guarantee.
Want your litigation counsel to review this?
We'll send the full memo and the Data Processing Agreement, with the retention schedule and sub-processor terms, in one pack.