Part of the AI Regulation News hub.
California DRE tells licensees that using an AI tool to perform licensed activity may be the same violation as handing that work to an unlicensed assistant
This one is five months old and we are covering it now because nobody covered it then. It is not news. It is the licensing regulator explaining, in writing, how it intends to read existing supervision and discipline law when a licensee points an AI tool at licensed work.
Bottom line: An advisory is not a regulation and did not go through rulemaking. It binds nobody by itself. What binds are the statutes and DRE regulations it construes, all of which were already in force. Dated March 17, 2026, still posted and still operative as of August 24, 2026.
Who this affects: California real estate brokers and salespersons, designated officers of corporate brokerages, residential property managers, brokerage compliance and risk officers, MLO endorsement holders, and California real estate counsel.
Issue date: March 17, 2026. The DRE assigns no advisory number and files it by publication date. No comment period, no effective date, no deadline of its own. One statute it discusses, Business and Professions Code section 10140.8, the DRE dates to January 1, 2026.
What changed: No law changed. What is new is the DRE saying on the record that a broker supervisory obligation extends to the tools, and that AI-assisted licensed activity can be analysed the way unlicensed-assistant activity is analysed.
Analysis: Our four earlier California and multistate real estate items were all image or video disclosure statutes, and the AVM item was a federal quality-control rule. This is a different lever. Disclosure statutes ask what you told the consumer. The licensing regulator asks who performed the act and who was supervising, which is the theory that reaches a licence rather than a listing.
Primary sources: DRE Licensee Advisory, AI in California Real Estate (March 17, 2026) · DRE Licensee Alerts and Advisories index · DRE guide to unlicensed assistants (cited in the advisory)
- Instrument (EN)
- Licensee Advisory: Artificial Intelligence in California Real Estate - Opportunities, Risks, and Compliance Considerations for Licensees
- Authority
- California Department of Real Estate (DRE)
- Jurisdiction
- California, United States
- Status
- Published and posted on the DRE Licensee Alerts and Advisories index under 2026
- Bindingness
- Not binding of its own force. No rulemaking, no advisory number, no penalty attaches to the advisory itself. It construes Bus. and Prof. Code sections 10131, 10140.6, 10140.8, 10159.2, 10176 and 10177(h), sections 6125 to 6126, DRE Regulations sections 2725 and 2780, Government Code sections 12955 to 12956.1, the Holden Act, and Civil Code sections 1798.100 to 1798.199.
- Issue date / next deadline
- Issued March 17, 2026. No deadline set by the advisory.
- Population reached
- The DRE states in its own footer that it oversees the licensure of approximately 434,000 licensees.
- Reported enforcement
- We found no DRE disciplinary action turning on AI use. See the last section.
- Primary source
- https://dre.ca.gov/Licensees/Advisories/Advisory_2026_03_17_AI_in_California_Real_Estate.html
An advisory is not a rule, and this one is not new
The DRE published this on March 17, 2026. It sat on the Licensee Alerts and Advisories index between an impersonation-scam alert from January and a scammer alert from April, and it did not get picked up. Five months later the page is still there and the law it construes has not moved, which is why it is worth writing up now. Nothing about it is a development.
Get the classification right before you act on it. This document did not go through rulemaking. It has no advisory number, no comment period, and no operative date of its own. It is the licensing regulator stating how it reads obligations that already existed, which is a different thing from creating one.
That distinction matters in a practical way. If a DRE accusation ever cites this advisory, it will not be charging a violation of the advisory. It will be charging Business and Professions Code section 10177(h) or section 10176 or DRE Regulations section 2725, and offering the advisory as evidence that the licensee had notice of how the Department reads them. Notice is the whole function of a document like this.
The unlicensed-assistant theory is the sharp end
Most AI guidance from professional regulators lands on process advice: review the output, keep a human in the loop, understand the tool. This advisory does that too, in a best-practices list at the end. But it also does something the disclosure statutes do not, which is to reach for the licence itself.
The Department writes: "Using AI tools to conduct licensed activity may be equivalent to asking an unlicensed assistant to do licensed activity, which is a violation of California real estate law." It then points licensees at its own long-standing guide to what unlicensed assistants may and may not do, which is a deliberate move. California already has a settled body of practice on where the line sits between clerical support and licensed activity under Business and Professions Code section 10131. The advisory is saying, in effect, that the existing map applies to the new tool.
Read the modal verb. The sentence says may be equivalent, not is equivalent. The Department has not announced a per se rule that any AI-assisted act is unlicensed practice, and it would be overreading the text to say it has. What it has done is put a recognised, chargeable theory on the table and tell 434,000 licensees where the analogy comes from. Our own reading, offered as inference rather than as the Department speaking: the analogy does real work because the unlicensed-assistant framework is fact-driven and already familiar to DRE investigators, so it needs no new authority to run.
Supervision follows the tool, and the vendor is not on the hook
The advisory grounds broker responsibility in sections 10177(h) and 10159.2 and in Regulation section 2725, then extends it in one line: a broker supervisory obligation extends to the tools used to conduct licensed or unlicensed activities, including AI-powered software.
The allocation of blame is stated flatly. "If an AI tool generates inaccurate information, misleading advertising, or improper communications with consumers, responsibility under current law rests with the licensee and their responsible broker, not the technology provider." Note the phrase under current law. The Department is describing the existing allocation, not creating one, and it is not purporting to override whatever a vendor contract says between the parties. It is saying the vendor is not who the DRE will be looking at.
On the review question the text is careful, and so should anyone quoting it be. The advisory says AI outputs "should be reviewed for accuracy and approved by the licensee before being relied upon in transactions or consumer communications." That is a should, phrased as a duty to use AI responsibly. Documentation is weaker still. The advisory says documentation of review, training and compliance practices may also help demonstrate good-faith efforts to comply with the law if questions arise, and lists Document Compliance Steps among its best practices. It does not say a documented review is required. Anyone telling California brokers the DRE now mandates an AI review log is reading in an obligation the page does not contain.
Consequences are listed, in the conditional. Failure to review AI output could subject a licensee to discipline for breaching fiduciary duties, negligence, substantial misrepresentation, or failing to make required disclosures, and a broker could face discipline for failing to reasonably supervise. Every one of those is an existing ground.
Advertising, altered images, and the pieces that are actually statutory
Three of the concrete items in the advisory rest on statute rather than exhortation, and those are the ones a compliance officer should separate out.
First, section 10176 already authorises discipline for substantial misrepresentation, false promises and deceptive practices, and the advisory applies it to AI-drafted listing descriptions, marketing emails and online advertisements. The instruction is independent verification of factual claims including property features, pricing, availability and potential uses. Use of an AI tool, the Department says, will not excuse the violation.
Second, section 10140.6 requires the licensee name, licence identification number and responsible broker identity on first-point-of-contact solicitation materials. The advisory simply says that requirement does not lapse because the content came out of a model. Worth knowing if your brokerage has chatbots or automated email sequences opening conversations with consumers.
Third, and this is where the advisory touches the ground our earlier coverage was on, section 10140.8 applies from January 1, 2026 to digitally altered images in real estate advertising. On the Department account it requires clear disclosure where an image has been modified in a way that changes the appearance of the property, including modification created or enhanced by AI, and requires the original unaltered image to be made available to consumers. The advisory adds that failure to review AI-generated images for compliance may result in violations even if the alteration was unintentional. One drafting note from the live page: the hyperlink attached to the section 10140.6 reference resolves to section 10176 rather than to 10140.6. Read the section numbers in the text, not the links.
Fair housing, tenant screening, privacy, and the practice-of-law line
The fair housing passage is the one property managers should read twice. The advisory names the Fair Employment and Housing Act at Government Code sections 12955 to 12956.1, the Holden Act on residential financing, and DRE Regulations section 2780, which prohibits advertisements indicating a preference, limitation or discrimination based on listed characteristics.
Four use cases are singled out for caution: targeted advertising, tenant screening, lead prioritisation and pricing recommendations. The Department says even neutral-appearing criteria may result in discriminatory outcomes or indirect discrimination, and that use of AI does not shield a licensee from liability under the real estate law, FEHA, the Holden Act or other fair housing and lending laws. Lead prioritisation is the interesting inclusion. A brokerage that would never write a discriminatory advertisement may be running a lead-scoring model it has never audited, and the advisory puts that squarely inside the same frame.
On data, the advisory routes licensees to the California Consumer Privacy Act at Civil Code sections 1798.100 to 1798.199 and tells them to establish before adoption what a tool collects, how long it retains it, whether it is shared with third parties or available to the general public, and what safeguards exist. It also says licensees should avoid inputting confidential or sensitive client information into public or unsecured AI platforms.
The last legal boundary is the unauthorised practice of law. AI tools that generate contract language, explain legal rights or interpret disclosure documents put a licensee near a line that is criminal, not merely disciplinary: sections 6125 to 6126. AI-generated explanations and document summaries, the advisory says, do not replace attorney advice, statutory disclosures, or the obligation to recommend that a consumer seek counsel.
What we did not verify
What we opened and read in full: the DRE Licensee Advisory at dre.ca.gov/Licensees/Advisories/Advisory_2026_03_17_AI_in_California_Real_Estate.html, which returned HTTP 200 on August 24, 2026 and carries the date line March 17, 2026 above the title; and the DRE Licensee Alerts and Advisories index, which lists it under 2026 with no advisory number, between the January 23 and April 27 items. Both quotations in this article were character-matched against the extracted page text. The approximately 434,000 licensee figure is the DRE own footer text on that page, not our estimate.
What we did not open: the underlying code sections themselves. We did not pull the current text of Business and Professions Code sections 10131, 10140.6, 10140.8, 10159.2, 10176, 10177(h) or 6125 to 6126, DRE Regulations sections 2725 or 2780, Government Code sections 12955 to 12956.1, the Holden Act, or the CCPA sections. Everything this article says about what those provisions require is the DRE characterisation of them, and we have marked it that way throughout. We also did not open the DRE unlicensed-assistant guide PDF, and we did not open the DRE enforcement actions database.
What we refuse to claim: that this advisory is a regulation, a rule, or new law, because it is none of those and went through no rulemaking. That any obligation in it began on March 17, 2026, because the obligations it describes predate it. That a documented AI review is required, because the text says such documentation may help demonstrate good faith, which is not the same thing. That the DRE has disciplined anyone over AI use. We searched for such an action and found none reported, and we did not query the enforcement database directly, so the honest statement is that we cannot confirm any, not that none exists. And we make no claim about how any other state licensing regulator would analyse the same facts.
Treat this as a notice document, not a new obligation. The compliance point that transfers is the allocation: when an AI tool produces a bad listing description, a misleading chatbot answer or an altered photo, the California licensing regulator looks at the licensee and the responsible broker, and the vendor is not in the frame. If you run a California brokerage, the cheap move is a written AI policy and a record of who reviewed what, which the advisory frames as good-faith evidence rather than as a requirement.
Source File
https://dre.ca.gov/Licensees/Advisories/Advisory_2026_03_17_AI_in_California_Real_Estate.html
Open the DRE advisory page and confirm four things. One, the date line reads March 17, 2026 and no advisory number appears anywhere on the page. Two, the sentence beginning "Using AI tools to conduct licensed activity" sits in the Broker Supervision and Licensee Responsibility subsection. Three, the Responsible Use section says documentation of review, training and compliance practices "may also help demonstrate good-faith efforts", with may and not must. Four, the footer states the DRE oversees approximately 434,000 licensees. Then open the Licensee Alerts and Advisories index and confirm the item is filed under 2026 by month, not by number.
Using AI tools to conduct licensed activity may be equivalent to asking an unlicensed assistant to do licensed activity, which is a violation of California real estate law. ยท California Department of Real Estate, Licensee Advisory, March 17, 2026
FAQ
Does this advisory create a new obligation for California licensees?
No. It went through no rulemaking, carries no advisory number and sets no deadline. It states how the DRE reads obligations already in the Business and Professions Code, the DRE regulations, and fair housing, fair lending and privacy law. The advisory itself binds nobody; the provisions it construes do.
Does the DRE require a documented review of AI-generated content?
Not on the text. The advisory says AI outputs should be reviewed for accuracy and approved by the licensee before being relied on, and separately that documentation of review, training and compliance practices may help demonstrate good-faith efforts if questions arise. Document Compliance Steps appears in a best-practices list. That is recommendation and evidentiary framing, not a stated requirement.
Can the technology vendor be held responsible instead of the brokerage?
Not by the DRE. The advisory states that where an AI tool generates inaccurate information, misleading advertising or improper consumer communications, responsibility under current law rests with the licensee and their responsible broker, not the technology provider. That describes regulatory exposure and says nothing about what a licensee might recover from a vendor under contract.
How is this different from the AI listing photo and video disclosure laws?
Those statutes regulate what you tell the consumer about an image or a video, and they attach to the advertisement. This advisory is the licensing regulator addressing who performed the licensed act and whether a broker supervised it, under supervision and discipline provisions such as sections 10177(h) and 10159.2 and Regulation section 2725. It reaches conduct and licences rather than labels. The advisory does discuss section 10140.8 on digitally altered images, so the two frames overlap on photographs.
Related briefings
Sponsored Training
Practical AI training for regulated professionals, built around verification, documentation and a defensible process. See the courses.