ChatGPT is flagging my account for suspicious activity but won't let me log out of devices

I'm hitting a wall with a "Suspicious activity detected" warning that's completely disrupting my workflow, and the weirdest part is that the built-in security tools to fix it are currently broken. The banner tells me someone else might be using my account and that I need to secure it to regain full feature access, but I've verified my active sessions and every single device listed is mine.

Why the "Log out of all devices" button is failing

The most frustrating part of this bug is the loop I'm stuck in. When I try to follow the security prompt to clear my sessions via Settings → Security → Active Sessions → Log out of all sessions → Log out of all devices, the system throws this error:

We were unable to process your request at this time. Please try again later.

On top of that, one of my own authorized devices is now showing a "Session status unavailable" message. It feels like a backend session-management bug because I'm being told to secure my account, yet the primary tool to do so is unresponsive.

Troubleshooting steps that didn't work

I've spent a significant amount of time trying to isolate this, but nothing has cleared the flag. For anyone else seeing this, here is exactly what I've already attempted:

• Switched between Chrome 153.0.8010.24 and other browsers
• Tested in incognito/private mode
• Wiped all cookies and cache
• Signed out and back in across multiple devices
• Verified no VPN, proxy, or iCloud Private Relay is active
• Confirmed no third-party clients or automation services are connected
• Enabled multi-factor authentication (MFA)
• Changed the password for the connected Google account

Despite all this, the warning persists. My environment is just an iPad running iPadOS 26.6.1 (23G83) and my personal phone. I don't share my account with anyone, so the "suspicious activity" flag is a total mystery.

Current status and timeline

This all started suddenly on September 7, 2026, around 3:30 PM Eastern Time. I've already opened a ticket with OpenAI Support and provided screenshots. They've escalated the case to a specialist, but the turnaround time is "in the coming days," which doesn't help when I'm in the middle of long-form writing projects and can't rely on the tool.

It seems like either a massive false-positive or a glitch in how they're tracking session states. I'm curious if others are seeing the "Session status unavailable" error specifically, as that feels like a technical clue that this is a platform-side issue rather than a security breach on my end.