Can a law firm use an AI receptionist?
In Oregon, yes, with conditions. State Bar Formal Opinion 2026-208, approved in February 2026, answers “Yes, qualified.” Lawyers must take reasonable steps to understand the tool, supervise it, be careful that it doesn’t give legal advice or mislead callers, and protect confidential information. The opinion also calls clear disclaimers important and prompt lawyer review prudent. Other states have their own rules, so check yours.
What Oregon’s ethics opinion says
Oregon State Bar Formal Opinion 2026-208, “Chatbots and AI Agent Communications,” covers software that interacts “through text or voice.” The points that matter most for a phone line:
- Competence (RPC 1.1). Lawyers must understand how the tool stores information and responds, and revisit that as it changes.
- Supervision (RPC 5.3). The lawyer is generally responsible for its conduct, and must ensure outside vendors perform competently and protect confidential information.
- Confidentiality (RPC 1.6, 1.18(b)). Before using it, lawyers must confirm that what it handles is reasonably secure, which may include encryption and keeping client information out of AI training.
- Accidental attorney-client relationships. In Oregon the question turns on the client’s reasonable belief. A bot that sounds like a lawyer could create one, so clear disclaimers matter.
- Honesty (RPC 7.1). It must not mislead callers about expertise, experience or past results. The opinion notes other states have treated a bot that doesn’t identify itself as potentially misleading.
- No legal advice. A bot giving legal advice could implicate the lawyer in unauthorized practice.
- Vulnerable callers (RPC 7.3(a)). A bot may miss an impaired emotional or mental state that a person would notice, so lawyers should review these conversations and interview each prospective client before forming a relationship.
- Prospective client duties (RPC 1.18). Information gathered can create conflicts, so it should go into the firm’s conflict system promptly.
- AI agents. Tools able to act on their own carry more risk, so autonomous actions should be subject to lawyer approval.
This is our reading for planning purposes and isn’t legal advice. The responsibility stays with the lawyer. We also have a longer guide to the opinion.
How we design an intake line around those duties
| Duty | How the line handles it |
|---|---|
| Disclaimers | The greeting says it’s an automated assistant, that it can’t give legal advice, and that calling doesn’t make the caller a client |
| No legal advice | It’s instructed to collect facts only. Questions like “do I have a case?” get a defined answer and go to the firm |
| No misleading claims | It’s set up to speak only from firm-approved information, with nothing about results or experience unless you approve it |
| Lawyer review | Each interaction notifies a responsible lawyer or trained staff member, with a transcript to review promptly |
| Conflicts | Adverse party names and contact details are collected and sent to your conflict process, with no substantive detail beyond what you choose |
| Limits on autonomous steps | It has no way to send engagement letters. Only steps you approve in advance, such as scheduling a consultation, happen without a lawyer |
What AI intake can cover
- Practice area screening. Is this something you handle? If not, it can say so politely and offer your referral information.
- Basic facts. Name, contact details, the type of matter, the other parties and relevant dates, following your intake script.
- Urgent matters. Calls about an upcoming court date or filing deadline go straight to a person by the route you set.
- Existing clients. It’s instructed to take a message for the responsible attorney without discussing the matter.
- Consultations. It can schedule one, and records go to your practice management software, such as Clio, where your account allows it.
Personal injury and family firms should think hardest about how a line handles a distressed caller. Our default is to keep intake short and get a person involved quickly.
Vendor due diligence
The opinion says due diligence should reach past the company you contract with, to the other software underneath it. A voice agent typically relies on several providers, for phone service, speech recognition, the language model and voice.
For a line we build, we’ll list the services call data passes through and point you to each provider’s published terms on storage, encryption and training use. Some providers offer stricter settings or different terms, and those can affect the choice of tools and the price. You decide if it meets your obligations.
Common questions from law firms
Is an AI receptionist ethical for an Oregon law firm?
It can be. Oregon State Bar Formal Opinion 2026-208 permits chatbots and AI agents for marketing and intake if lawyers maintain competence in the technology, implement safeguards and meet their professional obligations. The firm remains responsible for what the tool says.
Could an AI intake call create an attorney-client relationship?
The opinion warns that it could, because in Oregon the test turns on the client’s reasonable belief. That’s why clear disclaimers matter, along with prompt review by a lawyer or trained staff.
Can the AI tell a caller whether they have a case?
No. It should collect information and leave any assessment to a lawyer. Providing legal advice is one of the risks the opinion specifically raises.
Does it work with Clio?
Clio offers ways to connect outside tools, and intake details can often flow into it. What’s possible depends on your plan and setup, so we confirm that before quoting.
We’re not in Oregon. Does this still apply?
The opinion applies to Oregon lawyers, but other bars have addressed similar issues, including Florida and New York City, both cited in the opinion. Check your own jurisdiction’s guidance. Tell us the requirements you’ve identified and we’ll say whether we can meet them.