AI privacy for union organizers: what your employer, the AI company, and a court can see

The Morse teamPublished November 6, 20269 min read

AI privacy for union organizers depends on whose account and whose device you type on: an employer can read the chats in a work AI account and may capture what you type on a work laptop, the AI company keeps what you type into a personal account, and the NLRB or a court can ask for records in a labor dispute. Federal labor law bars employers from spying on union activity, but none of these tools checks why a record is being kept before it keeps it.

We read the National Labor Relations Act and the NLRB's own pages on employee rights, its General Counsel memos from 2022 to 2026, its 2019 and 2023 decisions on work email and work rules, and the admin documentation from OpenAI, Microsoft, and Google, on November 6, 2026. This page covers privacy only, and it isn't legal advice: a union representative or a labor lawyer can tell you what applies at your workplace.

What can your employer's AI tools record about organizing?

Whatever you type into them, and on a work laptop possibly what you type into a personal AI account too. A work AI account gives its admins your chats, and a managed laptop can capture prompts sent to personal AI sites; our guide to what a work laptop shows your employer goes through each admin console. Three more details come from the vendors' pages, read November 6, 2026:

• Locking a chat doesn't hide it. OpenAI says "Messages from locked chats in ChatGPT Enterprise and Edu workspaces are included in conversation message logs" that the company's compliance tools receive (OpenAI).
• Copilot keeps a copy where you can't see it. Microsoft 365 Copilot prompts and responses are copied to "a hidden folder in the mailbox of the user," where compliance staff can search them with eDiscovery tools (Microsoft Learn, updated June 25, 2026).
• The list of watched apps is long. The same Microsoft page lists ChatGPT, Google Gemini, the consumer Microsoft Copilot, and DeepSeek as "Other AI apps," whose prompts and responses a company keeps once it turns on a collection policy that captures content.

Is it legal for your employer to monitor union activity?

Spying on it is not. Section 7 of the National Labor Relations Act gives employees "the right to self-organization, to form, join, or assist labor organizations," and Section 8(a)(1) makes it an unfair labor practice for an employer "to interfere with, restrain, or coerce employees" in using that right (NLRB). The NLRB's list of what employers may not do includes "Spy on employees' union activities" and "Create the impression that you are spying on employees' union activities" (NLRB).

The NLRB defines spying narrowly. It "means doing something out of the ordinary to observe the activity," and "Seeing open union activity in workplace areas frequented by supervisors is not 'spying'" (NLRB). A retention policy set up for the whole company copies every prompt, organizing or not, before anyone decides to look. Searching those prompts is a separate step. Microsoft's Communication Compliance can flag prompts typed into Copilot and other AI apps, matches Copilot prompts against custom keyword lists, and its setup guide suggests reviewing 100% of generative AI interactions when a company first starts looking at them (Microsoft Learn, updated June 22, 2026). To Microsoft's credit, usernames there are "pseudonymized by default," and an admin must opt investigators in, but a reviewer who opens a flagged prompt sees the whole message. Whether a given search is "out of the ordinary" is for the NLRB to decide, case by case and after the fact.

The guidance that would have gone further is gone, though a January 2024 post that still ranks for this topic was written before the rescission. On October 31, 2022, General Counsel Jennifer Abruzzo issued memo GC 23-02, which described employers that "monitor employees who work on computers" with "keyloggers and software that takes screenshots, webcam photos, or audio recordings," and said she would urge the Board to find such practices "presumptively violated Section 8(a)(1)" where they could chill organizing (NLRB). On February 14, 2025, Acting General Counsel William Cowen rescinded it outright in memo GC 25-05 (NLRB). The NLRB's own memo list now marks it "Rescinded 2/14/2025 by Memorandum GC 25-05" (NLRB). That post, from a Georgetown Law journal, presents the memo as the General Counsel's current position (Georgetown Law).

Can you organize on work devices and work email?

On its own equipment, your employer can usually bar it. In Caesars Entertainment (December 2019), the Board ruled that employers "have the right to control the use of their equipment, including their email and other IT systems," so long as they "do not discriminate against union or other protected concerted communications," with an exception where work email is "the only reasonable means" for employees to talk to each other during non-working time (NLRB). In September 2024, the Board applied that rule to personal use of employer IT and "declined to reconsider" it (NLRB).

Work rules in general fall under Stericycle (August 2, 2023), which presumes a rule unlawful if it has "a reasonable tendency to chill employees from exercising their rights," unless the employer shows a "legitimate and substantial business interest" that no "more narrowly tailored rule" could serve (NLRB). That standard may not last. On August 26, 2026, General Counsel Crystal Carey listed among her priorities "Arguing to overturn Stericycle" in a pending case against Honeywell (NLRB). By the time any ruling comes, a prompt typed on a managed laptop may already sit in the company's archive.

A personal phone you enrolled for work email shows your employer far less: Microsoft says Intune never sees a device's "Calling and web browsing history" or "Email and text messages" (Microsoft Learn). Ask IT what your profile covers all the same, and keep organizing out of the work apps on it.

What does the AI company keep from a personal account?

Everything you type, until you delete it. As of November 6, 2026, OpenAI says personal ChatGPT chats "remain saved to your account until you delete them," and deleted chats are scheduled for permanent deletion "within 30 days," unless OpenAI "must retain it longer for security or legal obligations" (OpenAI). On Free, Plus, and Pro, "data sharing is enabled by default," so chats can train OpenAI's models until you turn off Improve the model for everyone, and a Temporary Chat "may be retained for up to 30 days for safety purposes" (OpenAI).

Training and retention apply to every user, but they weigh more for an organizer, because an organizing chat holds other people: coworkers' names, who has signed a card, and who is wavering. We compare what the big AI apps keep, and which settings change it, in how to use AI privately.

Can your AI chats be subpoenaed in a labor dispute?

Yes, by more than one route. The NLRB itself can demand evidence: Section 11 of the Act lets the Board "issue to such party subpoenas" on a party's application, and it must revoke one that doesn't relate to the matter or doesn't describe the evidence "with sufficient particularity" (NLRB). In a lawsuit, an employer can seek your records in discovery, though the Board limits how far it can dig into protected activity. In Guess?, Inc. (June 30, 2003), it held that questioning about Section 7 activity must be relevant, must not have an illegal objective, and must serve an interest that outweighs "the employees' confidentiality interests under Section 7 of the Act" (NLRB).

Requests to the AI company follow its own rules. OpenAI's law enforcement policy, effective January 1, 2026, discloses user content only "in response to a valid warrant or equivalent," gives notice "before disclosing the information, unless prohibited by law," and is "not intended for requests for information or service of legal process by OpenAI users or other civil or criminal litigants" (OpenAI). In a civil labor case, then, the request is more likely to come to you, for the history you can export. We cover how that works in can AI chats be subpoenaed.

Who could learn what from an organizer's AI chats

Each row is built from the sources above, read November 6, 2026.

Is it safe to use AI for union organizing?

On your own device and your own account, mostly yes, if you write about the work without naming the people. Drafting a leaflet, reading a contract clause, or asking what Section 7 covers puts little at risk. Names, sign-up counts, and who said what in a meeting identify coworkers who never chose to be in your chat, so keep those out of every AI tool. Never use a work account, a work laptop, or the work browser on your phone for any of it.

How Morse keeps organizing chats off servers your employer can reach

Every route above starts with a copy of your chat held by someone else: your employer's compliance archive, a mailbox folder you can't see, or an AI company's servers under your name. An employer can search its own archive without asking anyone, and a court can order a company to produce what it stores. Morse is built so the copy that matters, your chat history, stays on your device.

• E2EE, for anything that names coworkers or plans. Your browser checks the secure enclave's proof that it's genuine before anything leaves your device, then seals the message so only that enclave can read it. Morse's servers relay ciphertext they can't read, and the reply is decrypted only in your browser. The enclaves are run by third-party providers, not by Morse, and the guarantee depends on the enclave hardware and attestation working as designed.
• Private, for drafting and everyday questions. New chats start here, with open-weight models like DeepSeek, Kimi, and GLM, plus Grok. Morse has zero data retention agreements with its inference, search, and other providers, so they keep nothing. Where a provider confirms zero retention on each reply, Morse checks it and refuses a reply that doesn't.
• Anonymous, for GPT, Claude, or Gemini without the provider knowing it's you. Our proxy removes your account, email, and IP address and sends the message under Morse's own account. The provider may keep the message for a limited time to check for abuse, under its own policy, so the advice about names applies here most.

In every mode, your chat history is stored in your browser, encrypted, and not on Morse's servers, which don't keep your messages or log what you write. It locks when you sign out, and unlocking takes two keys, one of which never leaves your device. 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). Keep your work email out of any account you use for organizing. On a managed laptop, monitoring software sees the page before anything is sealed, so use Morse on a device that's yours. Clearing your browser or losing the device deletes your chats for good, and Morse can't recover them, so keep a copy of any draft you need. On a shared network, the network still sees the site's name, as we explain in what a WiFi owner can see.

Send a message from the Morse home page on your own device without an account, or start free with no card needed. E2EE chats come with Pro, at $19 a month.