MorseSign inStart freeOpen Morse

Can landlords use ChatGPT? Where applicant SSNs, credit reports, and tenant emails go

The Morse teamPublished 8 min read

Yes, landlords can use ChatGPT for listings, replies to tenants, and making sense of a lease clause, but an applicant's Social Security number, credit report, pay stubs, and bank statements don't belong in a consumer chatbot. As of November 2026, the FTC tells landlords to securely dispose of a screening report "and any information you gathered from it" (FTC), and ChatGPT keeps what you paste until you delete it.

This isn't legal advice. On November 16, 2026, we read the FTC's guide for landlords and its tenant page, the Fair Credit Reporting Act and the FTC's disposal rule, HUD's 2024 screening guidance (in an Internet Archive copy, since HUD no longer posts it), HUD's two 2026 withdrawal notices, the CFPB's May 2025 withdrawal notice, Massachusetts and New York rules on Social Security numbers, and the terms, privacy pages, and usage policies of OpenAI, Anthropic, and Google. The AI companies earn credit here. Anthropic's usage policy names rental applications outright, and Google tells you plainly what not to type.

Can landlords put applicant information into ChatGPT?

Not the screening report or anything taken from it. The FTC's guide Using Consumer Reports: What Landlords Need to Know says you can only get a report with a permissible purpose, and you "must certify" to the screening company "that you will use the report only for housing purposes." The Fair Credit Reporting Act backs that: no one may "use or obtain a consumer report for any purpose unless" the purpose is allowed and certified (15 U.S.C. 1681b(f)).

A tenant report holds more than a score. The FTC's tenant page, dated August 2025, lists "your name, Social Security number, date of birth, and current and past addresses," criminal records, and sometimes "a recommendation or score."

The FTC's disposal rule covers "any record about an individual, whether in paper, electronic, or other form, that is a consumer report or is derived from a consumer report," and anyone holding one for a business purpose must take "reasonable measures to protect against unauthorized access" when getting rid of it (16 CFR 682). Its examples are shredding paper and erasing electronic media so the information "cannot practicably be read or reconstructed." A shredder handles the printout. Once the report is pasted into ChatGPT, the copy on OpenAI's servers runs on OpenAI's retention clock instead. OpenAI's US privacy policy says it keeps conversations until you delete them and then removes them "within 30 days," unless the content "has already been de-identified and disassociated from your account" because you let it train on your chats.

What do ChatGPT, Claude, and Gemini say about other people's data?

All three put the applicant's privacy on you. We read each company's own pages on November 16, 2026.

ChatGPT (OpenAI)Claude (Anthropic)Gemini (Google)
Your promise about what you paste"You represent and warrant that you have all rights, licenses, and permissions needed to provide Input" (Terms of Use, effective January 1, 2026)"You are responsible for ensuring you have the necessary rights, permissions, or authority to do so" (privacy policy, effective September 10, 2026)Google will "process personal data about others that you've given Gemini Apps" (Gemini privacy hub)
Training and reviewMay use content "to train the models that power ChatGPT" unless you opt outUses chats for training "unless you opt out," and flagged chats even if you doSome chats go to human reviewers, so it asks you not to enter "confidential information that you wouldn't want a reviewer to see"
How long it can stayUntil you delete it, then within 30 daysDeleted chats leave the back end "within 30 days"Reviewed chats kept "up to three years," even after you delete your activity
Decisions about rental applicationsNo "automation of high-stakes actions" in housing "without human review" (usage policies)A qualified person must review, and the applicant must be told AI was used (usage policy)No "automated decisions that have a material detrimental impact on individual rights without human supervision" in housing (prohibited use policy)

Anthropic also keeps a separate non-user privacy policy, effective July 8, 2026, for "individuals who are not registered users" whose data is processed through its services. An applicant whose pay stub you paste into Claude is one of them, and that applicant never agreed to Anthropic's terms.

Can you use AI to screen tenants or rank applications?

Only with a person making the call, under the AI companies' own rules. Anthropic's usage policy, effective November 12, 2026, lists housing as a high-risk area and defines it as "approving, denying, or ranking a rental or purchase application; terminating a tenancy or initiating eviction or foreclosure." It covers "anyone who submits inputs" to Anthropic's products, including people using its apps like Claude.ai, not only businesses. A qualified person "must meaningfully review" the recommendation, and the applicant "must be clearly told that AI was used to produce it."

OpenAI's usage policies, in the version effective November 20, 2026, bar "automation of high-stakes actions in sensitive areas without human review," with housing on the list. They also rule out "evaluation or classification of individuals based on their social behavior, personal traits, or biometric data." Google's prohibited use policy, last modified December 17, 2024, names housing among the high-risk domains.

The credit law adds its own step. If a screening report plays any part in turning someone down, the FTC says the adverse action notice "is required even if information in the consumer report wasn't the primary reason for the decision" (FTC).

Is HUD's 2024 guidance on AI tenant screening still in effect?

It's no longer on HUD's website, and no notice we found says it was withdrawn. HUD's Office of Fair Housing and Equal Opportunity issued the guidance on April 29, 2024. It said a landlord or screening company can break the Fair Housing Act by screening on a protected trait or a proxy for one "even if the decision for how to screen applicants is made in whole or in part by an automated system, including a system using machine learning or another form of AI." The PDF at hud.gov now returns "not found," and Internet Archive captures show that error since March 6, 2025 (archived copy).

Yet neither of HUD's 2026 withdrawal notices lists it. The April 6 notice withdrew the companion guidance on digital advertising, effective September 17, 2025 (Federal Register). The July 17 notice withdrew 13 Office of General Counsel documents, including the 2016 criminal records guidance, effective September 25, 2025 (Federal Register). Both say the withdrawn documents "should not be relied upon." The CFPB, for its part, withdrew its January 2024 advisory opinion on background screening on May 12, 2025 (Federal Register).

The statute hasn't moved. HUD's April notice says conduct that doesn't comply with the Act's text will "continue to be subject to enforcement," and an applicant can still sue within two years (42 U.S.C. 3613).

Do state laws limit where an applicant's SSN can go?

Some do, and they reach small landlords. In Massachusetts, anyone who "receives, stores, maintains, processes, or otherwise has access to" a resident's name with a Social Security, driver's license, or account number needs a written security program (201 CMR 17.02 and 17.03). That program must oversee service providers by "requiring such third-party service providers by contract" to protect the data, and must encrypt records "that will travel across public networks" (17.04). Ask whether your AI app has signed anything like that before an applicant's Social Security number goes into it.

New York's General Business Law section 399-ddd tells anyone holding Social Security numbers for business to limit access to "a legitimate or necessary purpose" and to "protect the confidentiality of such number" (New York Senate). Other states have their own rules, so check yours before an application leaves your screening portal.

What can landlords safely ask AI to do?

Most rental writing needs nothing from the applicant file.

  • Listings. Describe the unit: rooms, light, parking, pet policy. The Fair Housing Act bars any rental ad "that indicates any preference, limitation, or discrimination" on a protected basis (42 U.S.C. 3604(c)), so leave out who you picture living there.
  • Tenant emails. Paste the problem and drop the name, the unit number, and the phone number. A request for a disability accommodation can carry a diagnosis, so describe it in general terms or write that reply yourself.
  • Lease review. Paste the clause you don't follow, with names, rent, and the address removed. Our guide to private AI for legal questions covers what else to leave out when the question turns legal.
  • Screening. Keep reports, pay stubs, and bank statements in the screening company's portal. Ask the AI how a process usually works, such as what an adverse action notice contains, with no applicant in the question. Our page on files you upload to ChatGPT explains how long uploads stay.

How Morse keeps your rental chats private

Each rule above comes back to a readable copy of someone else's data, held by an AI company you never signed a contract with, on terms that make you vouch for it. Your chat history is stored in your browser, encrypted, not on Morse's servers, and Morse doesn't train AI on your content, sell your data, or show ads.

The applicant file still stays where it is. Social Security numbers, credit reports, and income documents belong in your screening portal and property records, not in any chat. Morse is for the writing around them:

  • E2EE, for a tenant problem you need to think through in your own words. A late-rent dispute or a repair fight, with names left out, belongs in a chat only the verified enclave can read. Your browser checks the secure enclave's cryptographic proof before anything leaves your device, then encrypts the message so only that enclave can read it, and our servers relay ciphertext they can't read. The enclaves are run by third-party providers, not by Morse, and the guarantee rests on the enclave hardware and attestation working as designed. E2EE runs open-weight models like GLM, Kimi, and DeepSeek, with no web search or image generation.
  • Private, for listing copy, reply templates, and stripped lease clauses. It's the default for a new chat, on open-weight models plus xAI's Grok. Morse has zero data retention agreements with its inference, search, and other providers, so they've agreed to keep nothing. Where a provider confirms zero retention on each reply, Morse checks it and refuses a reply that doesn't (what zero data retention means).
  • Anonymous, to ask GPT, Claude, or Gemini how a rule usually works. Our proxy removes your account, email, and IP address, so the provider sees the question without knowing it's you. It may keep it for a limited time to check for abuse, under its own policy, so keep tenants out of these chats entirely.

Because we don't store chats, we usually have very little to give authorities (privacy policy). A Temporary chat is kept in memory only, which suits a notice you'll copy into your own records and close. Clearing your browser or losing the device deletes your chats, and Morse can't recover them, so keep every lease and notice in your property files.

Try a chat from the Morse home page without an account, or start free. End-to-end encrypted chats come with Pro, at $19 a month.

Common questions

Can I use ChatGPT to write a rental listing?

Yes, if the prompt describes the unit and not the people you hope will rent it. The Fair Housing Act bars any rental ad that indicates a preference based on a protected trait, so give the AI the property's features and read the result before you post.

Do I have to tell applicants I used AI?

If you use Claude to approve, deny, or rank rental applications, Anthropic's usage policy says the applicant must be clearly told that AI was used, and a qualified person has to review the recommendation first. OpenAI and Google bar fully automated housing decisions without human review.

Does turning off training in ChatGPT make it safe for applicant data?

No. It stops training, but OpenAI still keeps your conversations until you delete them, then removes them within 30 days, so an applicant's details still sit on its servers in the meantime.

Is HUD's 2024 tenant screening guidance still in effect?

As of November 16, 2026, it's gone from HUD's website, though neither of HUD's 2026 Federal Register notices withdrawing fair housing guidance lists it by name. The Fair Housing Act itself still applies.

Ask what you actually want to ask.

80+ models, characters, and images, with every chat encrypted and kept on your device. Free to start.

Start a private chat

Read next