Your agents are using AI faster than your brokerage can govern it.

Why this jumped to the top of the agenda

On June 22, 2026, the National Association of Realtors held a Regulatory Issues Forum on AI regulation as part of its legislative meetings. It was chaired by Cyndee Haydon, alongside a director from the White House Office of Science and Technology Policy, and reported by GlobeNewswire. The room was standing only.

Pay attention to who was on that stage. A national trade body does not put a White House technology official in front of a packed forum for a topic that can wait. The presence of both, in the same room, on AI regulation, tells you where this is heading. Rules are coming, and supervision is the part that will land on broker-owners first.

You do not need to predict the exact regulation to act. The direction is clear enough. The brokerages that get ahead of this will be governing agent AI use on their own terms before anyone forces them to. The ones that wait will be doing it under a deadline, or after a claim.

The gap that lands on your desk

Here is the uncomfortable part, and it comes straight from the industry's own risk people. Risk and Insurance reports that agents are adopting AI faster than their firms can govern it. Read that as a liability statement, because that is what it is.

The broker of record is responsible for supervising the agents under the license. That is not new. What is new is that your agents are now using tools you have never reviewed, that touch client data, fair housing language, and disclosure, often without telling you. The activity is real. The supervision is missing. And when supervision is missing, the exposure does not sit with the agent who clicked the button. It rolls up to you.

This is the trap of moving fast. Every agent who finds a clever AI workflow feels like a win for the brokerage. But each ungoverned workflow is a small open door. You are accountable for all of them at once, and most of them are invisible to you right now.

Three places the risk is real right now

Abstract risk is easy to ignore. So here are the concrete ones that show up in brokerages today.

• Client data in chatbots. An agent pastes a buyer's financial details, a seller's situation, or a full contract into a public AI tool to get help. That data may now sit on a third party's servers, outside any agreement you have, with privacy and confidentiality implications you never approved.
• Listing descriptions with no review. AI generates fast, fluent copy. It also invents features, misstates square footage, and produces language that can stray into fair housing trouble. If it goes live unreviewed under your brokerage's name, the misrepresentation exposure is yours.
• AI voice callers with no policy. Agents are using automated AI callers to work leads. Consent, disclosure, and do-not-call rules apply to those calls, and a tool running unsupervised under your license can generate violations at a scale a human never could.

None of these are exotic. They are happening in ordinary brokerages this quarter. The common thread is that each one is an agent making a reasonable-seeming choice with no policy to guide it and no review to catch it.

A brokerage AI policy you can adapt

The answer is not a ban. A ban gets ignored, pushes the activity underground, and leaves you with the same liability and less visibility. The answer is a written policy plus a light supervision routine. Here is the frame.

• A short, written AI use policy. One or two pages, plain language, that every agent signs. What tools are approved, what client data may never be entered into a public tool, and which outputs require broker review before they go to a client or go live.
• A client-data rule with teeth. The single most important line: no client personal or financial information into an unapproved AI tool, ever. Make it unambiguous and make the consequence clear.
• A review gate on outbound content. Listing descriptions, marketing copy, and any client-facing AI text get a human check before publication. It can be quick, but it has to exist.
• A rule for AI voice and outreach tools. Either approve a specific tool with compliant consent and disclosure built in, or prohibit them until you have. No agent-by-agent freelancing on automated calls.
• A named owner and a regular check-in. One person responsible for the policy, and a recurring agenda item where AI use, new tools, and any near-misses get surfaced. Supervision is a routine, not a memo.

The goal is not to slow your agents down. It is to make the AI use visible and governed so the speed is yours, not a hidden exposure.

Keep the agent tactics, govern the brokerage

One clarification so this lands right. This is a broker-owner topic, not an agent-productivity topic. How an individual agent writes a better listing or works leads with AI is a separate skill, and a good one. We cover the agent-level workflows in how real estate agents use Claude and the wider picture in how real estate runs on AI.

Your job is different. You are not optimizing one agent's output. You are governing a whole roster of people using tools you did not choose, under a license you are responsible for. The agents can and should keep their tactics. You set the guardrails inside which those tactics are safe for the brokerage. Both things are true at once, and the brokerage that does both pulls ahead.

The skill under the policy

A policy document on its own is paper. The real skill is judgment: knowing which AI uses are fine, which need a review gate, and which are flatly off limits in a regulated brokerage, and being able to supervise that without becoming the bottleneck for every agent.

That skill outlasts any single tool or rule change. If you want the structured version built for broker-owners and managing brokers, The Leveraged Real Estate Series teaches it from the ground up, and the two minute course quiz will point you to the right program for your brokerage.