ChatGPT unusual activity detected: the 6 causes OpenAI names, the fix for each, and how to spot an account review

What does ChatGPT unusual activity detected mean?

It means ChatGPT paused your session after OpenAI's systems saw traffic that looked automated. OpenAI's troubleshooting page puts it in one line: "This message is shown when our systems detect traffic that may resemble automated or unusual behavior." A deactivation is a different thing, and it arrives by email. The exact wording depends on where you hit it, as of October 2026:

All four share one gap. OpenAI doesn't say how long they last: not the troubleshooting page, not the suspicious activity article, not the network article with a section on the iPhone message, and not the account deactivation article. Screenshot the reference code after the message (one forum poster's read 89c80d9aa9cb5b49-VIE). OpenAI's alert article asks for any request, ray or device IDs when you contact support.

Why does ChatGPT say unusual activity has been detected from your device?

ChatGPT says it because one of six triggers matched, and OpenAI's help pages and support team point at all six. Most come down to your IP address, your account or your app. Find your row and try that fix first.

The VPN row is the one OpenAI repeats. Its troubleshooting page says to "Disable VPNs or proxies that might route your traffic through flagged IPs." The alert article adds Private Relay, the Safari feature whose second relay gives you a temporary IP address. On an iPhone it sits under Settings, your name, iCloud, Private Relay, per Apple's setup steps. Shared accounts break the rules as well as tripping the check. OpenAI's Terms of Use, effective 1 January 2026, say "You may not share your account credentials or make your account available to anyone else."

Which fix actually clears it? We read every post in OpenAI's forum thread on the device message on 11 October 2026. That's 49 posts from June 2024 to September 2026. Twenty posters said a step cleared their own error, and a "worked for me too" reply counts under the post it answered.

Six people tried steps and stayed blocked. No single fix is certain. One that worked appears in no help page. On 7 May 2026 a poster found the phone's clock set 5 minutes ahead. Switching it to automatic ended both the unusual activity message and a TLS error, after OpenAI Support's own list had failed.

How do you fix unusual activity on the ChatGPT iPhone or Android app?

Start with the app, not your account. OpenAI lists iOS steps for this exact message in its network article, and OpenAI Support posted the same list in the thread on 18 February 2026. The steps below follow it, with the clock fix added.

1. Copy anything unsent, then sign out of the ChatGPT app and sign back in.
2. Update the app from the App Store or Google Play. If nothing changes, uninstall it and reinstall it.
3. Turn off any VPN, and Private Relay if you use Safari.
4. Set the phone's date and time to automatic.
5. Open chatgpt.com in the phone's browser. If the browser works and the app doesn't, your account is fine and the app is the problem.
6. Contact support from help.openai.com with a screenshot, the reference code and the time.

OpenAI's list ends somewhere unusual for a chat app. It suggests setting the phone's DNS to 8.8.8.8 and 8.8.4.4, then opening register.appattest.apple.com in Safari and sending support a screenshot of what loads. App Attest is Apple's check that an app is genuine: Apple's servers "certify that the key belongs to a valid instance of your app", per Apple's developer docs. Google's Play Integrity API does the same job on Android, and a May 2025 poster's error read "preauth playintegrity verification failed". Our reading, which OpenAI hasn't confirmed: on a phone, "from your device" can mean the app couldn't prove it's a genuine copy.

The Mac app has its own fixes. A November 2024 post that deletes the app's folder in ~/Library/Application Support and then reinstalls drew three replies saying it worked. Two more posters force quit the cfprefsd process, and one deleted only the two newest files in the app's conversations folder. Quit the app before deleting anything.

How do you fix it on chatgpt.com in a browser?

Use the five troubleshooting steps from OpenAI's suspicious activity article, in its order. They're free, and none deletes your saved chats.

1. Sign out of all devices, then sign back in.
2. Clear cookies and cache for chatgpt.com.
3. Try a private window, another browser or another device.
4. Turn off any VPN, proxy, Private Relay, extension or automation tool.
5. Stay on one trusted device and network while you test.

Step 4 is the long one. OpenAI's list reads "Disable any VPN, proxy, Private Relay, browser extensions, or automation tools." Older forum reports add remote desktop software: a January 2024 poster said closing it stopped the error.

How long does the ChatGPT unusual activity block last?

OpenAI doesn't say, and forum reports run from hours to weeks. Blogs that promise 5 to 10 minutes aren't quoting OpenAI. These are the durations posters gave, with the date of each post:

Posters tied two waves to OpenAI's side. In the December 2023 thread, the first poster said support traced it to an issue hitting some enterprise users and fixed it with a rollback. In May 2024 the web thread's opener tied it to incidents open on OpenAI's status page. If the message follows you to every device and network, check status.openai.com before you change anything else.

Is it a temporary block or an account review?

It's a temporary block when the message sits inside ChatGPT and says try again later. It's an account review when OpenAI emails you or the sign-in screen says the account is deactivated. OpenAI's deactivation article, updated September 2026, lists suspected compromise as one reason it acts: "we may temporarily suspend it to protect you and prevent misuse."

The alert article is calm about the first case: "These alerts are often cautionary and do not necessarily indicate wrongdoing." The deactivation article lists what turns a flag into a review. It names usage policy violations, shared accounts or API keys, "Circumventing security or access restrictions", and warnings left unanswered. Scripts belong in that list too. The Terms of Use bar anyone who would "Automatically or programmatically extract data or Output".

Is unusual activity the same as a usage limit or something went wrong?

No. Unusual activity is about who's sending, and a limit message is about how much. A usage cap names a limit or a reset time, which the ChatGPT limit reached guide covers. "Something went wrong" is the general error with server or network causes, and our fix for that error runs seven steps in order. The unusual activity fixes target your address, account and app instead. If ChatGPT wants proof of age, that's a separate check: why ChatGPT asks for ID.

If you need answers while the block lasts, use another app, and what to use when ChatGPT is down lists the options. Whizi, which publishes this page, is one. It sends prompts through OpenRouter rather than your chatgpt.com login, so a flag on that session doesn't follow you. Starter, at $15.99 a month, carries ChatGPT Luna, Claude Haiku and Whizi AI. Where it loses: Whizi has its own rate limits, 10 messages a minute and 60 an hour per user, plus 60 requests a minute per IP address.