Is there a private ChatGPT you can pay for? What Plus, Business, and the rest change
The Morse teamPublished 12 min read
Not a fully private one: OpenAI's personal plans don't make ChatGPT private, and its business plans do it only in part. As of November 20, 2026, ChatGPT Plus ($20 a month) and ChatGPT Pro ($100, $200, or $500 a month) buy more usage, not more privacy (2, 3). OpenAI trains on chats from both plans unless you opt out, keeps them on its servers until you delete them, and files every one under an account with your name on it. The plan OpenAI sells with training off by default is ChatGPT Business, from $25 a seat a month, or $20 billed annually (10), but its workspace admins "can view, access, export, and delete" your conversations (11).
The facts behind that answer, read on OpenAI's own pages today:
- Training. On Free, Plus, and Pro, "data sharing is enabled by default" (4). A thumbs up or down lets OpenAI use "the entire conversation" for training even after you opt out (5).
- History. OpenAI keeps saved chats on your account until you delete them. Once you do, it schedules each one for permanent deletion "within 30 days," with exceptions for security and legal duties (7).
- Identity. OpenAI's privacy policy lists what your account holds, from your "name, contact information, account credentials" to your "date of birth" and payment details, and says the log data OpenAI collects "includes your Internet Protocol address" (9).
- Temporary Chat. It stays out of your history and out of training, but OpenAI "may keep a copy of a temporary chat for up to 30 days for safety purposes" (6).
- Business and Enterprise. OpenAI doesn't train on either by default (4). Enterprise admins set how long chats are kept, and OpenAI publishes no Enterprise price (1, 11).
We read OpenAI's pricing page, its US privacy policy (updated September 10, 2026), its enterprise privacy page (updated January 8, 2026), and nine of its help articles on November 20, 2026, with the pricing and privacy pages of DuckDuckGo, Proton, Brave, and Maple. The numbers in parentheses point to the Sources list at the end. OpenAI earns credit for one thing most companies leave out: its consumer data article lists the exact reasons its staff may read your chats.
Morse is a private AI chat app with GPT, Claude, Gemini, DeepSeek, and uncensored models in one place, encrypted so even Morse can't read your chats. Morse Pro costs $19 a month, with unlimited messages and no daily cap. In Anonymous mode, every message goes through the Morse Enclave, which sends it on under Morse's name, so OpenAI reads your message to answer it without learning your name, email, account, or IP address. No AI provider trains on what you send through Morse.
Does paying for ChatGPT Plus or Pro make it private?
No. On OpenAI's pricing page, every personal plan from Free to Pro carries the line "Content is used to train our models," each one marked "Opt-out available" (1). The Plus help article puts it in its own words: "OpenAI may use conversations to improve model performance and safety" (2). The money goes to models and usage. Plus is "$20/month (billed monthly)," with no annual billing (2), and the three Pro tiers cost $100, $200, and $500 a month, with Ultrafast modes only on the top one (3).
Paying doesn't change who can open your chats either. OpenAI says "a limited number of authorized OpenAI personnel," along with trusted service providers, may access your content for four reasons: investigating abuse or a security incident, supporting you when you ask for help, handling legal matters, and "to improve model performance (unless you have opted out)" (8). Your chats are stored "on OpenAI systems and our trusted service providers' systems in the US and around the world," and the article ends with a line worth taking literally: "Please do not enter sensitive information that you would not want reviewed or used." (8)
Plus does remove one thing. OpenAI says "Plus, Pro, Business, Enterprise, and Edu accounts will not have ads" (24), and its privacy policy collects "ads history and interests" for Free and Go users only (9). Our full read of the consumer side, from memory to shared links, is in is ChatGPT private.
How do you stop ChatGPT Plus from training on your chats?
Turn off one setting, then leave the thumbs alone. On the web, open your account menu, choose Settings, then Data controls, select Improve the model for everyone, turn it off, and select Done (4). OpenAI says you keep your history when you do, and the choice follows your account to every device you sign in on (4). Selecting Do not train on my content in OpenAI's Privacy Portal does the same job, and "Either option is sufficient" (5).
Three gaps stay open after you opt out:
- Feedback. Rating a reply can still send the whole conversation to training (5).
- Trusted Access sessions. During Daybreak sessions on a personal account, OpenAI "may use content from your interactions to improve our models even when 'Improve the model for everyone' is off" (4). How ChatGPT's training settings work is in does ChatGPT train on your data.
- The past. The switch covers what comes next: "we won't use your new conversations to improve our models" (5).
The history of these rules, and what each switch has covered over time, is in does ChatGPT train on your data.
Does Temporary Chat make ChatGPT private?
It makes a chat disappear from your sidebar, not from OpenAI. A temporary chat stays out of your history, doesn't create or update memories, and isn't used to train OpenAI's models while it stays temporary (6). OpenAI still "may keep a copy of a temporary chat for up to 30 days for safety purposes," and if you choose Personalized before the first message, the chat can draw on your existing memories, custom instructions, and plugins (6). Save it, and it becomes a regular chat under your normal training and retention settings (6).
It is also sent from your signed-in account, with the log data the privacy policy describes (9). Ways to use ChatGPT with less of your identity attached are in how to use ChatGPT anonymously.
What do ChatGPT Business and Enterprise change?
Mainly the training default and who holds the keys to your chats. OpenAI says plainly that "OpenAI won't train on your workspace's data" for Business (10), and the same default covers Enterprise (4). In exchange, your chats sit in a workspace an administrator controls.
| As of November 20, 2026 | ChatGPT Plus | ChatGPT Business | ChatGPT Enterprise |
|---|---|---|---|
| Price | $20 a month, billed monthly only (2) | Standard seat $25 a user a month, or $20 billed annually, and Premium seat $125, or $100 billed annually (10) | Not published, "contact sales" (1) |
| Seats to start | One person | "1 paid Standard or Premium seat" (10), though the pricing page says "starting at 2 users" (1) | Not published |
| Trains on your chats | Yes, with an opt-out (4) | Not by default (4, 10) | Not by default (4) |
| Who in your organization can read them | No one else | Admins "can view, access, export, and delete end user conversations" (11) | Admins can "access an audit log of conversations" through the Compliance API (11) |
| Who at OpenAI can read them | Staff and service providers, for four listed reasons (8) | Employees for "engineering support, investigating potential platform abuse, and legal compliance," and contractors "solely to review for abuse and misuse" (11) | Employees "only" to resolve incidents, recover chats "with your explicit permission," or where the law requires (11) |
| How long chats are kept | Until you delete them, then removal within 30 days (7) | Admins "can control how long your data is retained," and deleted chats are removed within 30 days (11) | Admins control retention, and deleted chats are removed within 30 days (11) |
| Zero data retention | No | Only through "a contracted offering instead of self-serve ChatGPT Business" (10) | Through sales |
Business is the cheapest way to get ChatGPT with training off from the start, and OpenAI's help center says one person can open a workspace with a single seat (10). Its pricing page, read the same day, still says Business starts at two users (1), so check the checkout before you count on one. Buying it for yourself makes you the admin, and your chats still live on OpenAI's systems, open to the staff and contractors in the table. A Business member can't export their own data either: OpenAI sends that request to the workspace owner (4). What an employer can see in a workspace someone else runs is in is ChatGPT Business private.
What would a private ChatGPT actually need?
Four things, and each OpenAI plan misses at least two. Price tags and plan names make every option sound private, so these are the tests we used for the table below:
- No training on what you send, by default, with no feedback button that reopens the door.
- No stored history the company can read. If the chats sit on its servers in readable form, staff access rules, breaches, and court orders all reach them.
- No identity tied to your prompts. The company running the model shouldn't receive your name, email, account, or IP address with each message.
- A way to check, so the first three rest on something more than a promise.
Business and Enterprise pass the first. None of OpenAI's plans passes the second or third, because every one stores your chats on OpenAI's systems under an account with your name on it (7, 8, 11).
What are the best alternatives to ChatGPT Plus for privacy?
The paid options people weigh against Plus are Morse, ChatGPT Business, the DuckDuckGo subscription for Duck.ai, Proton's Lumo Plus, Brave's Leo Premium, and Maple, and they differ most on two points: whether you get OpenAI's and Anthropic's models, and who can read your history. Each row comes from that company's own pages, read on November 20, 2026.
| As of November 20, 2026 | Price | GPT and Claude models | Trains on your chats | Where your history lives | What the model's operator learns about you |
|---|---|---|---|---|---|
| Morse Pro | $19 a month, or $180 a year | GPT, Claude, and Gemini, with DeepSeek, Grok, and uncensored models | No, and no AI provider trains on what you send | On your device, encrypted | In Anonymous mode, the provider never learns your name, email, account, or IP address |
| ChatGPT Business | $25 a seat a month, or $20 billed annually (10) | GPT only (1) | Not by default (10) | OpenAI's servers, readable by workspace admins (11) | Your workspace account |
| DuckDuckGo Plus or Pro (Duck.ai) | $9.99 or $19.99 a month, $99.99 or $199.99 a year (12) | GPT-5.6 Terra and Claude Sonnet 5.5 on Plus, plus Claude Opus 5.5 and GPT-5.6 Sol on Pro (14) | No, by agreement with providers (15) | On your device, or encrypted Sync & Backup (15) | Prompts arrive with IP address and personal metadata removed (15) |
| Proton Lumo Plus | $12.99 a month, or $119.88 a year (16) | None: open models such as Qwen 3.5 and GLM 5.3 (17) | No (17) | On your device, synced with zero-access encryption (17) | Proton runs the models and reads messages in cleartext to answer (18) |
| Brave Leo Premium | $14.99 a month, or $149.99 a year (19) | GPT 5.6 Terra, Claude Sonnet 5.5, and Claude Opus 5.5, among others (20) | No (20) | Only on your device (20) | Brave says usage "is unlinkable with subscription information" (20) |
| Maple Pro | $20 a month (21) | None: open models such as Kimi, GLM, and GPT-OSS (21) | "No, by design" (22) | Stored encrypted (23) | Messages decrypted "only inside a hardware-isolated enclave" (23) |
A few details the table can't hold:
- Duck.ai comes inside a four-part subscription with a VPN, Personal Information Removal, and Identity Theft Restoration (12), and Pro adds higher reasoning effort and "2x higher usage limits than the Plus plan" (13). DuckDuckGo deserves credit for writing its provider terms into its policy: providers delete prompts "at most within 30 days, with limited exceptions for safety and legal compliance" (15).
- Lumo is open source and runs only on servers Proton controls, with "no logs of what you ask" (17). Proton is direct about the trade: "we need to provide the LLM with cleartext messages" (18).
- Leo is built into the Brave browser, and Brave says every model, Claude included, runs on its own infrastructure with "no data sharing between Brave and the original model creators" (20).
- Maple lets you run the enclave check in your own browser, and its proof page says what the check doesn't prove: "You do not attest the model's enclave directly" (23).
Side-by-side reviews with more detail are in ChatGPT alternatives and Duck.ai vs Lumo vs Brave Leo.
How Morse gives you GPT without OpenAI knowing it's you
Every gap above comes from one design: your chats sit on the model company's servers, under an account carrying your name, email, and payment details. Paying OpenAI more moves the training switch and adds an admin, and the chats stay where they were. Morse is built so they never land there.
Training has no switch to find, because Morse doesn't train AI on your content and no AI provider trains on what you send through Morse. Your history is stored on your device, encrypted, and our servers keep no log of your messages, so there's no copy for staff to review. Your identity stays with us: the Morse Enclave never receives your account, email, or IP address. You can check that the enclave runs our published, open-source code on genuine AWS Nitro hardware with npx @askmorse/verify inference.trymorse.ai (how to verify Morse).
You choose who can see each message, chat by chat:
- Anonymous, to use GPT, Claude, or Gemini without the provider knowing it's you. The enclave sends your message under Morse's own account. OpenAI reads it to answer, may keep it for a limited time to check for abuse, and never learns who sent it.
- E2EE, for anything no one else should read. Your browser checks the enclave's hardware attestation, then encrypts the message so only that verified enclave can decrypt it. Our servers relay it without being able to read it. E2EE chats come with Morse Pro.
- Private, for open-weight models on hosts that keep nothing. The host reads your message only to answer it, and its terms with us bar it from keeping any of it.
In every mode, your chats are stored on your device, encrypted, and unlocking them takes a key from your browser and one from our server at sign-in. Morse doesn't sell your data or show ads. Because your chats are stored on your device, Morse has no copy of them to give authorities. Keep a backup from Settings > Data, since your device holds the copy. The full detail of who sees what is in our privacy policy.
Try a chat from the Morse home page without an account, or see Morse's plans: Morse Pro is $19 a month, with unlimited messages, no daily cap, and E2EE chats.
Sources
Read on November 20, 2026.
- OpenAI, Plans and Pricing
- OpenAI, What is ChatGPT Plus?
- OpenAI, About ChatGPT Pro tiers
- OpenAI, Data controls in ChatGPT
- OpenAI, How your data is used to improve model performance
- OpenAI, Temporary chat in ChatGPT
- OpenAI, Deleting and archiving chats in ChatGPT
- OpenAI, How OpenAI handles data in consumer services
- OpenAI, US privacy policy, updated September 10, 2026
- OpenAI, ChatGPT Business overview
- OpenAI, Enterprise privacy at OpenAI, updated January 8, 2026
- DuckDuckGo, How much does DuckDuckGo cost?
- DuckDuckGo, Plus and Pro plan differences
- DuckDuckGo, What AI chat models are available?
- DuckDuckGo, Duck.ai Privacy Policy and Terms of Service
- Proton, Lumo pricing and plans
- Proton, Lumo privacy
- Proton, Lumo privacy and security model, August 4, 2025
- Brave, Brave premium
- Brave Help Center, What are the differences between Leo's AI Models?, updated October 6, 2026
- Maple, Maple pricing
- Maple, Private AI chat and work
- Maple, How Maple works
- OpenAI, Ads in ChatGPT
This isn't legal advice. Prices and policies change, so check each company's page before you buy.
Common questions
Does ChatGPT Plus train on my chats?
Yes, unless you opt out. OpenAI says that on Free, Plus, and Pro, "data sharing is enabled by default," and you turn it off under Settings > Data controls > Improve the model for everyone.
Is ChatGPT Pro more private than Plus?
No. OpenAI's pricing page gives both the same training line, "Content is used to train our models," marked "Opt-out available." The $100, $200, and $500 Pro tiers buy more usage and faster models, not different data handling.
Can I buy ChatGPT Business just for myself?
OpenAI's Business help article says you can start a workspace with one paid seat, at $25 a month or $20 a month billed annually. Its pricing page still says Business starts "at 2 users," so check the checkout before you plan on one seat.
Is ChatGPT Enterprise available to individuals?
Not on any page we read. OpenAI publishes no Enterprise price and sends buyers to its sales team, and it describes Enterprise as a plan for organizations.
Does opting out of training delete what OpenAI already has?
No. OpenAI says that after you opt out, "we won't use your new conversations to improve our models," and turning the setting off "does not delete or hide saved chats." Delete chats separately if you want them gone.
Do I still get ads on ChatGPT Plus?
No. OpenAI says Plus, Pro, Business, Enterprise, and Edu accounts will not have ads, which may appear on the Free and Go plans. Its privacy policy collects ads history and interests only for Free and Go users.
Is Duck.ai's paid plan private?
DuckDuckGo says it removes your IP address and other personal metadata before a prompt reaches the model provider, and that providers agree not to train on it and to delete it within 30 days at most, with exceptions for safety and legal compliance. The subscription costs $9.99 a month for Plus or $19.99 for Pro.
Can I run a private ChatGPT on my own computer?
Not ChatGPT itself, which runs only on OpenAI's servers. Local apps run open-weight models on your own hardware, and our guide to whether Ollama is private covers what stays on your machine.
Can Morse read my chats?
No. Morse keeps your chat history on your device, encrypted, and Morse's servers keep no log of your messages.
Does OpenAI know it's me when I use GPT through Morse?
No. In Anonymous mode, the Morse Enclave sends your message on under Morse's own account, so OpenAI reads it to answer without learning your name, email, account, or IP address. OpenAI may keep it for a limited time to check for abuse, and doesn't train on it.