Don’t miss our latest video – Inside AI Episode 7 with Summer Goralik, a leading expert on AI and real estate compliance.

AI is getting better at doing things on behalf of real estate agents, and that is exactly where the legal questions begin. Once an AI system starts communicating with consumers, interpreting information, giving advice or taking action on your behalf, it may be moving into activity that requires a real estate license. The problem is simple: AI cannot get one.

That creates a problem the industry has barely begun to address. As AI moves from generating content to becoming your digital surrogate, who is responsible when the technology crosses a legal line? That was at the heart of my latest Inside AI conversation with Summer Goralik, a leading real estate compliance consultant.

Summer brings a perspective few people in real estate can match. She is a California real estate broker and compliance consultant who previously investigated hundreds of complaints for the California Department of Real Estate and later worked with the Orange County District Attorney’s Real Estate Fraud Unit. I first interviewed her during a panel I moderated at Inman Connect San Diego, and I knew immediately there was a much bigger AI conversation to have.

We had it, and one idea Summer shared may be the simplest way I have heard to think about the legal limits of agentic AI. Before letting AI do something for you, ask what your state allows an unlicensed assistant to do.

The unlicensed assistant test

When I wrote about this issue recently for our weekly REAL AI newsletter and podcast, I use Summer’s unlicensed agent analogy to describe AI when used by a real estate agent today.

The problem gets much more serious with agentic AI. A generative AI tool can draft an email and wait for you to review it. An AI agent can send the email, communicate with the consumer and potentially take additional actions without waiting for you.

Summer said she initially started going down the regulatory rabbit hole trying to figure out where all this fits. Then she found a much simpler way to look at it.

“What can an unlicensed person do?” she asked.

Her advice is to look at what your own state regulator says an unlicensed assistant can and cannot do. “What is your regulator telling you about an unlicensed assistant? Because that is your threshold,” Summer said.

The unlicensed assistance rule does not answer every future question about AI, and state laws may differ. But until regulators provide more AI-specific guidance – which she notes is desperately needed in almost every state – it gives brokers and agents something they can use today.

When AI starts talking, listen closely

The biggest concern comes when AI begins communicating directly with consumers, particularly when that communication moves beyond providing basic information.

“If you have an unlicensed person engaging with a consumer, you are playing with fire,” Summer said. She becomes even more concerned when a chatbot begins advising someone, because that can move the technology into activities traditionally performed by a licensed professional.

That distinction is becoming more important as AI companies race to build autonomous tools for real estate. The technology may be able to do something, but that does not automatically mean an agent or broker can legally turn that responsibility over to it.

Summer summed up the risk in a way that is hard to forget: “AI’s not going to get in trouble.” The licensee is the one with something to lose.

Waiting for the first AI case

Here is the uncomfortable part: We still need regulatory guidance that specifically addresses agentic AI in real estate.

Summer told me that when she spoke earlier this year to regulators from around the country at an ARELLO meeting, very few had issued AI advisories. Meanwhile, as she put it, agents are already “knee-deep in it.”

She believes the industry may wind up learning through enforcement. Many regulatory investigations begin with a consumer complaint, and Summer thinks eventually an AI mistake will create one significant enough to establish some of the boundaries everyone is looking for.

“It’s only going to take one complaint,” she said.

That is not the way anyone should want AI policy to develop. Regulators, brokerages and technology companies have an opportunity to provide guardrails before a consumer gets hurt and an agent becomes the test case.

The best second set of eyes ever?

What I like about Summer’s perspective is that this is not an anti-AI argument. In fact, one of the most interesting parts of our conversation was hearing how bullish she is about AI’s potential to improve compliance.

“I do think that AI can serve as the best second set of eyes ever,” she said.

Think about the upside of that for a moment. AI can review enormous amounts of information, flag inconsistencies, spot something a person missed and give an agent or broker another layer of protection. When used correctly, it could make agents more efficient and potentially more compliant.

But it does not change who owns the final decision. Summer’s point throughout our conversation was that licensed activity, judgment and responsibility cannot simply be handed to the machine because the technology has become capable enough to take over the task.

Keep your hands on the wheel

Summer has compared AI to having a Ferrari in the driveway. It is incredibly powerful and can take you places faster than you imagined, but the ability to go 200 mph does not mean you should.

That analogy works particularly well for agentic AI because the technology is accelerating much faster than regulation. The road rules around licensing, supervision and fiduciary duty are still there, even when the car suddenly can drive itself.

That is why this conversation with Summer is worth watching. We are entering an AI era where the most important question may no longer be, “Can AI do this?” It may be, “Should I allow AI to do this for me?”

The safest answer starts with something we have been preaching since the beginning of REAL AI: keep the human in the middle. Summer puts an even finer point on it: “The licensee has to remain the driver.”

Watch Episode 7 of Inside AI with The REAL AI Guy for my full conversation with Summer Goralik, leading real estate compliance consultant.

# # #

Kevin Hawkins, a partner with WAV Group, is editor and co-founder of REAL AI, real estate’s No. 1 weekly AI newsletter and companion podcast, and his Inside AI with The REAL AI Guy video interview series. Kevin has written more about AI than anyone in real estate and is the Amazon bestselling author of The REAL AI Guide for Real Estate Agents. Known throughout the industry as The REAL AI Guy, in the last 18 months, he has taught thousands of real estate agents in the US and Canada how to use AI in the best and safest ways.