AI privacy for domestic violence survivors: who can see your chats, and what tells them

The Morse teamPublished October 31, 202610 min read

AI privacy for domestic violence survivors depends on who else can open your accounts and devices. As of October 2026, ChatGPT, Claude, Gemini, Meta AI, and Alexa+ save what you type or say to your account, and anyone signed in to that account, on any phone, laptop, or speaker, can read it. Someone who monitors you can also learn from what changes, because logging out other devices, removing a contact, or wiping your history can show up on their side.

If you're in danger, call 911. The National Domestic Violence Hotline answers by phone at 1-800-799-7233 and by live chat at thehotline.org, 24 hours a day, and by text: send START to 88788. The Hotline's own site warns that "Internet usage can be monitored and is impossible to erase completely" (The Hotline), so if someone may be watching this phone or computer, it's safer to read this on a device they've never had access to.

We read the National Network to End Domestic Violence's Safety Net Project guides, the Hotline's tech safety pages, and the help pages of OpenAI, Anthropic, Google, Meta, Amazon, and Apple on October 31, 2026. This page is about privacy, and it isn't legal advice. An advocate can help you weigh each step below against your own situation.

Can my abuser see my ChatGPT history?

Yes, if they can open a device where you're signed in or they know your password, because your chats belong to the account, not the device. OpenAI says regular and archived chats "remain saved to your account until you delete them" (OpenAI), and warns that other devices where ChatGPT is open "may continue to show the chat title" of a chat you've just locked (OpenAI). We cover shared tablets, Family Sharing, and old share links in AI privacy in a divorce, and what a ChatGPT PIN hides in are ChatGPT Locked Chats encrypted.

An account holds more than its chat list:

• Memory. OpenAI says "You can also ask ChatGPT what it remembers about you" (OpenAI), and so can anyone else typing into your account. It adds that deleting a chat "does not necessarily delete a separate saved memory created from that chat." Anthropic says Claude's memory "is on by default for Free, Pro, Max, and Ultra plans," and shows what it holds under Settings, then Your Claude (Anthropic).
• Archives. Archiving a ChatGPT chat takes it off the sidebar, but OpenAI says "Archived chats remain searchable" (OpenAI).
• "Only visible to you." Meta says your Meta AI History "is only visible to you" (Meta), which means visible to whoever is signed in as you. Meta also says that when accounts share an Accounts Center, "Meta AI remembers details you share with it in chats across those accounts" (Meta).

How do I check who else is signed in to my AI accounts?

Each app keeps a list of the devices signed in to your account. Looking at the list changes nothing, but the person you sign out from it may notice.

• ChatGPT: Settings, then Security, then Active sessions. OpenAI says a row may show the device or browser, an "Approximate location," and the sign-in date and time, and the row marked CURRENT SESSION is you (OpenAI). It doesn't list connected apps or recently signed-out sessions.
• Claude: Settings, then Account, then Active sessions ("Where you're logged in" in the newer layout). Each row gives the browser, the operating system, an approximate location from the IP address, and when it was last active (Anthropic).
• Gemini: in your Google Account, Security & sign-in, then Manage all devices. Google's list covers devices signed in now or in the last few weeks (Google).
• Meta AI: in Accounts Center, Password and security, then Where you're logged in (Meta).
• Alexa: in the Alexa app, Alexa Privacy, then Review Alexa History, where you can filter "by device or profile" (Amazon).

OpenAI and Anthropic both show an approximate location on each session, which makes a device in another town easier to spot. If you find a device you don't recognize, write it down with the date before you act on it. The Safety Net Project's technology safety plan says documenting incidents "can show police or the court a pattern of behavior."

Will an abuser know if I change my AI settings?

Sometimes, yes. The same safety plan warns that "some abusive individuals may escalate their controlling and dangerous behavior if they feel they've lost access to the victim." Apple's personal safety guide says, "Before making changes or deleting information, consider potential impacts to your safety and privacy" (Apple).

We matched the common steps to what the other person might see, from the companies' own pages, read October 31, 2026.

OpenAI lets one adult per account be told if reviewers think "you may be talking about suicide in a way that signals a serious safety concern" (OpenAI). You can see who that is under Settings, then Trusted contact, or by asking ChatGPT, without changing anything.

The Hotline and the Safety Net Project both describe a quieter approach. Leave the monitored device and accounts looking as they always have, and do the private things somewhere else. The safety plan says "some survivors choose to use a safer device for certain interactions, but also keep using the monitored device as a way to collect evidence." The Hotline suggests keeping a monitored email account "active with non-critical emails in order to maintain appearances."

Is it safer to use AI on a different device?

Yes, on one the person has never been able to reach. The safety plan's first step is to use "a different device that the person hasn't had physical or remote access to in the past, and doesn't have access to now (like a computer at a library or a friend's phone)." The Hotline names libraries, shelters, workplaces, and computer technology centers, and says to "avoid using shared computers when researching things like travel plans, housing options, legal issues, and safety plans."

On that device, sign in to nothing the other person can open. The safety plan suggests a new email address "with a non-identifying username," and says "not to link these new accounts to any old accounts or numbers." NNEDV's Survivor's Guide to AI adds that where an AI app offers sign-in with Google or Facebook, "consider using a burner email or an account not linked to your real identity."

Each big app also has a chat that stays out of your history, with limits:

• ChatGPT's temporary chats "do not appear in your chat history," but OpenAI says they "may be retained for up to 30 days for safety purposes" (OpenAI).
• Claude's incognito chats stay out of chat search: "Claude won't pull information from incognito chats when searching previous conversations."
• Gemini's temporary chats are "retained with your account for 72 hours," and Google says that when you use Gemini signed out, "Your data is not associated with your Google account" (Google).

Can a smart speaker be used to monitor you?

It can. The safety plan lists "personal assistants (such as Google Home or Alexa)" among everyday devices used for monitoring, and says they "may be able to be secured by changing account settings or passwords," once you've planned for your safety and documented what happened.

What you say to one is kept with the account that owns it. Amazon says Alexa+ in a browser lets you "Find and keep previous conversations with Alexa, even those started with devices or on the Alexa app" (Amazon), so a question asked to the kitchen speaker can be read later on a laptop somewhere else. Google keeps Gemini for Home voice settings apart from Gemini's, in Google Home History (Google). We go further into Amazon's assistant in is Alexa+ private and Apple's in does Siri record your conversations.

Is it safe to tell an AI chatbot about abuse?

Not the details that identify you. The Survivor's Guide to AI says that unlike an advocate, "most AI tools store the things you tell them and keep logs of your conversations," and advises: "Think of AI as a public tool, not a private confidant." Its example is to write "my ex-partner did XYZ" instead of a name and an address. It also reminds you that an AI "won't proactively call for help if you're in danger."

The guide's main example of the risk has changed since it was written. It says "one court order forced an AI company to save all user chats, even those users thought they deleted." OpenAI says its duties under that order "ended on September 26, 2025," and that deleted chats and Temporary Chats are again deleted within 30 days, though it still stores "limited historical April–September 2025 user data" (OpenAI). The advice to keep details out still holds, because a court can order a company to produce any chat it stores, as our ChatGPT court order timeline shows. For the other personal things people tell ChatGPT, see is it safe to tell ChatGPT personal things.

How Morse keeps your chats off other people's devices

Every way in above runs through an account that keeps your history on a company's servers and shows it on any device signed in to it. Morse keeps your history in your browser instead, and a guest chat keeps none at all.

On a safer device, you don't need an account. A message sent from the Morse home page opens a guest chat that's kept in the tab's memory only and never saved, so there's no history to find afterwards. The visit can still show in that browser's history, so on a shared computer, clear that one site as the Hotline describes.

If you sign up, your chat history is stored only in your browser, encrypted, and Morse's servers don't keep messages or log message content. It's locked when you sign out, and unlocking it takes two keys: one that never leaves the device, and one our server gives you when you sign in, so a copy of the browser's data alone can't be read. You sign in with a code sent to your email, or with Google, so use an address only you can open. Clearing the browser deletes the history for good, and we can't recover it, so keep anything you're saving as a record somewhere only you can reach.

Each chat has its own privacy mode:

• E2EE, for writing down what happened. Your browser checks the secure enclave's proof that it's genuine, then encrypts the message so only that enclave can read it, and our servers relay text 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 chats come with Pro, at $19 a month.
• Private, for questions about housing, money, or a new job. It's the default for a new chat. Morse has zero data retention agreements with its inference, search, and other providers, so the hosts running open-weight models like DeepSeek, Kimi, and GLM keep nothing.
• Anonymous, for asking GPT, Claude, or Gemini without your name on it. Our proxy removes your account, email, and IP address, so the provider sees the message without knowing it's you. It may keep the message for a limited time to check for abuse, under its own policy, so the guide's rule of no names still applies.

Morse doesn't train AI on your chats, sell your data, or show ads, and because we don't store chats, we usually have very little to give authorities (privacy policy). A Temporary chat is kept in memory only.

• Someone who checks your phone or laptop: in ChatGPT or Claude, the history is there for anyone who opens the app on a signed-in device. On Morse, sign out when you're done, and the history locks.
• Someone who knows your AI password: check the sessions list from a safer device, and talk to an advocate before you log anyone out.
• A speaker in the house: ask it only what you'd say in front of the person, and type the rest on a safer device.
• The company or a court: keep names and places out, or start the chat in E2EE.

Try a guest chat from the Morse home page on a device that's yours alone, or start free with no card.