Key takeaways
- Free AI apps are fine for generic writing; the risk starts the moment client details, offers or financial facts are pasted in[2][3].
- Client confidentiality is the agent's obligation under the Code of Ethics, whatever the app's terms say[3].
- Every AI draft can be confidently wrong, so a licensee checks each fact against the MLS or the seller's paperwork before it goes out[4].
- Pay for a business tool when client data, a record of what was said, more than one user or the phone line is involved.
The free app is already in your office
Most brokerages never chose their first AI tool. An agent pasted a seller's notes into a free chat app late one night to get listing remarks written, the remarks came back clean, and by the end of the month half the office was doing the same. NAR's 2026 technology survey found that 41% of REALTORS use AI-generated content tools such as chat assistants, and among those using AI, 75% use it to write listing descriptions and 52% to draft emails and follow-ups[1].
The tools are not the problem. The gap is that nobody decided which tasks belong in a free consumer app and which do not. Across all industries, 44% of employees told Gallup their organization has started integrating AI, but only 22% said it had communicated a clear plan[5]. Clients have noticed the tools too: among sellers who hired an agent, 21% knew and another 36% suspected their agent was using AI[6].
This guide compares free consumer AI apps with paid business tools on the three things that matter to a brokerage: where client information goes, how often the output is wrong, and what the tool can do beyond writing. Price is the smallest part of it.
What a free account is paid with
A free consumer account runs on consumer terms. The person who signed up controls the settings, the provider can revise its terms, and nobody at the brokerage can see what was pasted in or delete it when the agent leaves. That is a reasonable deal for a personal recipe planner. It is a poor one for a seller's divorce timeline.
The Federal Trade Commission has spelled out the tension. Its technologists wrote that companies offering AI models have a business incentive to constantly ingest additional data, that customers may reveal sensitive or confidential information when they use these models, and that a company which breaks its privacy commitments, such as a promise not to train on customer data, may be liable under the laws the FTC enforces[2]. A month later the agency added that quietly switching to more permissive data practices, such as using data for AI training, through a surreptitious, retroactive change to the terms may be unfair or deceptive[7].
Read those two notices from the agent's side. The protection is only as good as the terms in force on the day of the paste, and on a free account the agent, not the brokerage, is the one who clicked accept. Paid business plans usually come with a written agreement, admin controls and a named place to ask what happens to the data. Ask for each of those in writing; a paid logo alone guarantees none of them.
Confidentiality is the agent's duty, not the app's
Standard of Practice 1-9 of the NAR Code of Ethics says REALTORS shall not knowingly reveal confidential information of clients, and that the duty to preserve it continues after the relationship ends[3]. State law defines what counts as confidential, but in practice the list is familiar: the seller's lowest acceptable price, why they are moving, a buyer's ceiling, a pre-approval letter, health or family details that explain a deadline.
Pasting any of that into a tool the brokerage does not control is a disclosure the client never agreed to, whatever the app later does with it. Clients are already wary. In Pew's 2023 survey, 73% of US adults said they have little or no control over the data companies collect about them, and among those who had heard of AI, 70% had little or no trust in companies to use it responsibly and 81% expected companies' AI use to lead to their information being used in ways they would not be comfortable with[8].
Generative AI can also expose information without a human sharing it on purpose. NIST's generative AI profile lists data privacy as a core risk and notes that models may leak, generate or correctly infer sensitive information about individuals[4]. The safest client detail is the one that never reaches a tool the brokerage cannot audit.
Reliability: confident and wrong
Free or paid, a general-purpose chat tool will sometimes invent facts. NIST calls this confabulation: confidently stated but erroneous or false content by which users may be misled[4]. In listing copy that looks like a finished basement the house does not have, a school assignment nobody checked, or an HOA that supposedly allows short-term rentals. The more polished the paragraph, the less likely a busy agent reads it twice.
Article 12 of the Code of Ethics asks REALTORS to be honest and truthful in their real estate communications and to present a true picture in their advertising and marketing[3]. An AI draft does not move that duty anywhere. The fix is procedural: write from a verified fact sheet, then check every number and feature against the MLS entry and the seller's documents before publishing. The step-by-step method is in drafting listing descriptions with AI safely.
Paid business tools reduce this risk only when they are built to answer from your own approved material instead of from general knowledge. That design question matters more than the subscription, and it is worth asking in every demo.
Free app or business tool, side by side
| Free consumer app | Paid business tool | |
|---|---|---|
| Whose terms apply | Consumer terms the agent accepted | A written agreement the brokerage signs |
| Who controls settings | Each agent, one account at a time | An admin, for every user at once |
| When an agent leaves | Their history leaves with them | Access is removed and records stay |
| Where answers come from | General knowledge, often unchecked | Your approved material, if built for it |
| Record of what was said | Whatever the agent saved | Logs the brokerage can review |
| Answers the office phone | No | Phone-answering tools do |
Business plans differ; confirm each right-hand item in the contract rather than assuming it.
| Task | Free app is fine | Use a business tool | Keeps a licensee in charge |
|---|---|---|---|
| Social post about the spring market | Yes, with no client names | Optional | Check any statistic |
| Listing remarks | Only from a fact sheet with no seller details | Better, with history kept | Verify every feature |
| Summarizing a purchase contract | No | Only with a data agreement | Licensee reads the original |
| Follow-up email to a named buyer | Generic template only | Yes, inside the CRM | Agent sends it |
| Answering calls to the office or sign line | Not possible | Yes, a phone-answering tool | Agent confirms showings and advice |
Where free tools stop: the phone
Every free AI tool an agent uses today works on text the agent types in. None of them picks up the sign rider number while the agent is inside a showing. That job needs a business tool, because it runs under the brokerage's name, talks to the public and has to leave a record.
An SI Employee is one example of that category. The brokerage forwards its office or sign number for the hours it chooses, the SI Employee answers listing questions from the listing sheet the brokerage approves, marks anything not on the sheet as not confirmed, takes showing requests for the agent to confirm, and leaves a recording, transcript and summary of each call (see what it does). The solo listing agent example shows how one agent with 12 listings would set it up, and preparing listing data covers the sheet itself.
To compare a free setup with a paid one honestly, put a number on the time the paid tool would take off people. The calculator below counts only agent time spent on routine calls. It leaves out the larger value of calls answered while everyone is busy, so treat the result as a floor.
| Input | Example |
|---|---|
| Calls a day across the team | 20 |
| Minutes per call | 5 |
| Share handled without an agent (%) | 40% |
| Agent time per hour | $27.59 |
| Calls a month | 434 |
| Staff hours on these calls | 36 |
| Hours handed off | 14 |
| Monthly value of that staff time | $399 |
How it is worked out: Calls a day × 21.7 working days × minutes per call ÷ 60 × share handled = hours; hours × hourly cost.
Assumptions
- Hourly value of agent time uses the BLS 2025 median pay for real estate brokers and sales agents, $27.59 an hour.[9]
- Example: 20 calls a day across a small team, 5 minutes each, with 40% routine enough (price, status, open house times, messages) to need no agent.
Example figures for a small team. Swap in a week of your own call log before deciding anything.
A free-tool rule agents will follow
Banning free tools outright rarely works; agents are independent and the apps are on their phones. A short rule that says what is allowed tends to stick better than a long one that says what is not. Put it in the brokerage's written policy (the brokerage AI use policy guide has a one-page template) and repeat it at the next sales meeting.
Strip identifiers before pasting
No client names, phone numbers, emails, street addresses of unlisted homes, loan figures or ages. Write "the seller" and "a 3-bed colonial in the north end" instead.
Never paste documents
Purchase agreements, disclosures, offers, inspection reports and pre-approval letters stay out of consumer apps entirely.
Switch off history and training where offered
Check the app's data settings on day one and again whenever the terms change, since they can[7].
A licensee reads every word
Nothing AI-drafted goes to a client, a portal or the MLS until an agent has checked it against the source.
Client data means a business tool
If the task needs real client details, it moves to a tool the brokerage has an agreement with, or it is done by hand.
State privacy law can add duties on top of this, particularly around what you tell consumers at collection and what your vendors may do with their data; consumer data privacy laws for brokerages walks through the CCPA and the states that followed it.
When paying is worth it
Most agents already carry a technology bill. In NAR's 2026 survey, 36% of REALTORS spend $50 to $250 a month on technology, 22% spend more than $500 and 18% spend less than $50, and 59% name cost as a challenge in adopting new technology[1]. Another subscription has to replace work, not add a login.
- Client information is part of the task. Free consumer terms are the wrong place for it.
- You need a record. If a client later disputes what was said or sent, the brokerage needs logs it controls.
- More than one person uses it. Shared work needs admin controls and a clean exit when someone leaves.
- It faces the public. Anything that speaks for the brokerage, on the phone or in chat, needs approved answers and a person accountable for them.
- It replaces measurable time. If you cannot name the hours or the calls it takes over, wait.
When those boxes are ticked, ask every vendor the same questions: is customer data used to train models, how long are inputs and recordings kept, can we delete them on request, who at the company can see them, and what happens to our data if we cancel. Get the answers in the contract. Then build the line item into the yearly plan; planning a real estate technology budget shows how to sort spend by function so the free tools and the paid ones each have a job.
Example: a five-agent team keeps a free chat app for social posts and market blurbs written from public data, pays for its CRM's built-in writing assistant for client emails, and forwards its sign numbers to a phone-answering tool. Nothing with a client's name ever touches the free app.
Frequently asked questions
Is it safe to paste a listing address into a free AI app?
A public, active listing address on its own is low risk, since it is already published. The risk is everything that tends to come with it: the seller's name, their reasons for moving, their bottom line. Under Standard of Practice 1-9 those details are the agent's to protect[3], so strip them before pasting.
Do free AI apps train on what agents type?
It depends on the provider, the plan and the settings, and terms can change. The FTC has warned that quietly changing terms to use data for AI training may be unfair or deceptive[7], which is also a reminder to recheck settings whenever terms are updated. Assume inputs may be kept unless a written agreement says otherwise.
Can agents use free AI to write listing descriptions?
Yes, if they write from a verified fact sheet with no confidential seller details and check every claim before it is published. Generative tools can state wrong facts with confidence[4], and the agent remains responsible for a true picture in advertising[3].
Should a brokerage pay for AI tools for every agent?
Not by default. Pay where the tool handles client information, needs records, is shared by several people or speaks to the public. For generic writing from public information, a free tool with a clear rule is usually enough.
Do clients care whether their agent uses AI?
Many already assume it: 21% of sellers who hired an agent knew and 36% suspected their agent was using AI tools[6]. What they care about is whether their information stays private and the facts are right.
References
National Association of REALTORS® (2026). REALTORS® Adopt Technology to Save Time and Improve the Client Experience, NAR Report Finds (2026 REALTORS® Technology Report) (opens in a new tab) ↑
Federal Trade Commission (2024). AI Companies: Uphold Your Privacy and Confidentiality Commitments (opens in a new tab) ↑
National Association of REALTORS® (2026). 2026 Code of Ethics and Standards of Practice (opens in a new tab) ↑
National Institute of Standards and Technology (2024). Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1) (opens in a new tab) ↑
Gallup (2025). AI Use at Work Has Nearly Doubled in Two Years (opens in a new tab) ↑
Zillow Research (2026). Sellers: Results from the Zillow Consumer Housing Trends Report 2026 (opens in a new tab) ↑
Federal Trade Commission (2024). AI (and other) Companies: Quietly Changing Your Terms of Service Could Be Unfair or Deceptive (opens in a new tab) ↑
Pew Research Center (2023). How Americans View Data Privacy (opens in a new tab) ↑
U.S. Bureau of Labor Statistics (2025). Occupational Outlook Handbook: Real Estate Brokers and Sales Agents (opens in a new tab) ↑