Real estate brokers have never carried this much responsibility for technology they did not select, approve, or even know their agents were using.
The AI holds no license. It passed no exam, carries no E&O coverage, and cannot be disciplined by a real estate commission. But the agents who use AI, and the brokers responsible for supervising those agents, can be held accountable. The license of the real estate agent and the brokerage hang in the balance of supervision, training, education, and contract terms.
Artificial intelligence has not created a new category of broker liability. It has taken one of the industry’s oldest obligations, the duty to supervise, and expanded it across every agent, every device, every application, and every hour of the day.
That makes AI the biggest broker-supervision risk in real estate history. Profoundly, the larger the brokerage, the greater the risk. One incident by one agent is excusable. Dozens or thousands of incidents is profoundly worse, particularly if it crosses state license boundaries.
The compliance perimeter is gone
For decades, brokers managed technology risk through procurement. They selected the CRM, transaction platform, website vendor, and marketing tools. Whatever technology agents used, the broker had usually reviewed the contract.
That perimeter no longer exists.
Agents are adopting general-purpose AI tools on their own. At the same time, vendors are quietly adding AI capabilities to products the brokerage already uses. A feature that was not present when the contract was signed may appear in the next software update.
The broker remains responsible for supervising licensed activity conducted under the brokerage.
This is not merely a technology problem. It is a license-law problem, and existing licensing and supervision rules may already provide the framework for addressing it.
When AI becomes licensed activity
Summer Goralik, a former California Department of Real Estate investigator and one of the industry’s leading compliance consultants, framed the issue clearly during a recent conversation with WAV Group’s Kevin Hawkins: Your AI agent cannot get a real estate license, but you can lose yours.
The risk increases when AI begins communicating with consumers, interpreting transaction information, giving advice, or acting on an agent’s behalf.
At that point, the question is no longer whether the technology is useful. The question is whether it is performing work that state law reserves for a licensed professional.
Regulators may not need a new AI statute to address that conduct. Existing licensing and supervision rules may already apply.
Treat AI like an unlicensed assistant
The message to agents should be direct: Treat AI as you would an unlicensed assistant. The machine can assist with the work. It cannot become the licensee.
Goralik’s warning should become a brokerage operating principle: AI output must be tested, challenged, and verified against authoritative sources and professional judgment before anyone relies on it.
When an agent’s license is at risk, the broker’s is too. Supervisory liability runs uphill. If AI-generated listing remarks create a fair housing violation, an automated system misrepresents a property, or a chatbot gives transaction advice without proper oversight, a regulator will ask the broker a simple question:
What did you do to prevent this?
Why AI creates an unprecedented supervision problem
Brokers have always supervised people. People can be trained, questioned, corrected, disciplined, and terminated. Their activities generally occur through company systems or leave records that management can review.
AI changes that equation.
One agent can use dozens of AI tools without notifying the brokerage. One tool can generate hundreds of communications. An automated system can repeat the same mistake across thousands of records before anyone notices. A model’s behavior can change after an update even when the brokerage has changed nothing.
The scale, speed, and invisibility are unprecedented. That is why AI is not simply another item for the compliance manual. It changes the scale of the broker’s supervisory obligation.
The Broker’s first move: require a signed AI policy
A policy buried in a manual is not enough. Every agent, affiliated licensee, team leader, and staff member should sign the policy and renew that acknowledgment annually. The technology will change faster than the document. The National Association of REALTORS(r) has published a document for Brokers and associations to consider for policy templates – https://www.nar.realtor/artificial-intelligence-real-estate
A defensible policy should:
- Treat AI as an unlicensed assistant subject to applicable state rules.
- Require a licensee to review and accept responsibility for every AI-generated output before it reaches a consumer, cooperating broker, or MLS.
- Prohibit users from entering confidential client information, transaction documents, or personally identifiable information into unapproved tools.
- Address fair housing, advertising accuracy, disclosure, and image manipulation in listing content.
- Identify approved tools and establish a process for reviewing new ones.
- Require agents to report material AI errors, unauthorized uses, consumer complaints, and possible violations.
- State that violations may be grounds for termination and may be reported when required by law.
The signature matters. It moves the broker’s position from “we hoped they understood” to “they were informed, acknowledged the rules, and agreed to follow them.”
But a signature is only the beginning.
A defensible supervisory system also requires training, internal reporting procedures, periodic review of AI-assisted work, documented corrective action, and evidence that the brokerage monitors compliance.
Kevin Hawkins offers a class that will allow brokers and associations to demonstrate that they are performing training and advising agents. Its affordable and effective.
The Broker’s Second move: find the AI already inside your company
Agents’ personal AI accounts are only half the exposure. The other half is embedded in the technology the brokerage already buys.
Vendors are adding AI features on rapid release cycles. A lead-response tool that once sent a template can now conduct a conversation. A website chatbot that once collected contact information may answer questions about a property, neighborhood, or offer process. A transaction platform may now claim to “review” contracts.
Many of these features arrive through software updates, not contract negotiations. Brokers may never have approved them or even know they are active.
The first step is an inventory. Brokerages need to identify:
- The AI tools agents and staff use independently.
- The AI features embedded in approved technology.
- The systems that communicate directly with consumers.
- The systems that access client, transaction, or MLS data.
- The systems that recommend or take actions without prior human approval.
- The records available to show what each system said and did.
If management cannot answer those questions, it does not yet understand its exposure. David Gumpper can perform an audit or help with your company’s audit.
Third broker move: amend every technology contract
Every technology agreement should require the provider to represent in writing that its product will not use AI to perform activities requiring a real estate license.
Stronger language should also require the vendor to:
- Disclose every AI-enabled feature.
- Explain what information the feature accesses, retains, and uses for training.
- Provide written notice before introducing new AI capabilities.
- Allow the brokerage to disable those capabilities.
- Maintain records of automated consumer communications and actions.
- Indemnify the brokerage if the vendor’s representations prove false.
Expect resistance. Some vendors may argue that they cannot make such a representation because licensing rules differ across states.
That response is revealing. The vendor may have introduced a product into a regulated industry without determining how those regulations apply.
The broker’s license should not serve as collateral for that uncertainty. If a vendor will not sign, its product deserves close scrutiny before renewal.
Victor Lund can audit your agreements and suggest language changes.
Why taking action matters
A brokerage license is the foundation beneath the entire enterprise. Commission revenue, agent count, and company value all depend on that license remaining in good standing.
A disciplinary action involving unsupervised AI, a machine-generated fair housing complaint, or an E&O claim complicated by the absence of AI controls could cost far more than years of compliance work.
The risk is not that AI will lose its license. It has no license to lose. The risk is that an agent will trust it too much, a broker will supervise it too little, and the brokerage will bear the consequences.
Strategic recommendation for the broker
Direct management to complete five actions within 90 days:
-
- Adopt a signed AI policy. Engage real estate counsel to draft a policy based on the state’s unlicensed-assistant and broker-supervision rules. Obtain a signed acknowledgment from every licensee and staff member, renewed annually with the independent contractor or employment agreement.
- Inventory AI across the brokerage. Identify both approved and unsanctioned tools used by agents and staff, along with every AI capability embedded in the company’s existing technology.
- Amend technology contracts. Require representations about licensed activity, disclosure of AI features, advance notice of changes, appropriate data protections, the ability to disable features, and indemnification. Make the language standard in new agreements and renewals.
- Confirm insurance coverage. Ask the company’s E&O carrier, in writing, how the policy treats claims arising from AI use, automated communications, data exposure, and the absence of human review. Address any gaps.
- Create accountability and escalation. Assign a named officer responsibility for AI compliance. Require agents to report material AI errors, unauthorized uses, consumer complaints, and possible violations. Track corrective actions and recurring patterns, and report quarterly to the board.
- Provide agent training
Supervision remains with the broker. The license at risk belongs to the brokerage. Brokers who act now will look prudent. Those who wait may become the case studies everyone else learns from.
Watch Kevin Hawkins’ conversation with Summer Goralik: Inside AI, Episode 7: “Your AI agent can’t get a real estate license. But you can lose yours.”
Read Goralik’s related analysis: “A compliance expert quizzed ChatGPT. Here’s what it got wrong.”
