"The encrypted content gAAA...5f0= could not be verified."

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

• #8121
• #8119

Powered by Codex Action

I am wondering if this is external since I have just started seeing this issue too

@Chargeuk Yes, but this didn’t happen to me before.

Thanks for reporting. We're aware of the problem, and the team is actively investigating.

• feedback ID 019b2885-07d0-7382-b9d8-2b3d5e4423ac

Does anyone know if downgrading Codex version helps? We need mitigation steps

Does anyone know if downgrading Codex version helps? We need mitigation steps

I noticed that it works good for small tasks, apparently when it needs to involve context and bigger tasks it brokes.

Also, tried to logout and login again, tried to clear the extension local storage and reinstall the extension, nothing changed same behavior always on big tasks.

This started about 30 min ago.

It seems to be something related to VS Code extension, since via codex webpage it is working.

Same error here, I don't use VS Code, codex-cli 0.73.0

019b28b7-fec6-74e2-8625-f25d8abf6d35

{
  "error": {
    "message": "The encrypted content gAAA...Mss= could not be verified.",
    "type": "invalid_request_error",
    "param": null,
    "code": "invalid_encrypted_content"
  }
}

By the way, I was able to resume one of the conversations with the same prompt, it produced the answer straightforward

Same issue in Codex Cli, each time I get this error I tell it to "continue". It will keep going for a while and then hit the error again. The error doesn't seem to wipe the context, it reports to have 50% context left. So I'm attempting to get it to spit out a summary or answer before giving up. I seems to be working, I'll also try to resume the converesation in the VS Code plugin interface to see if that helps.

• No such luck, multiple tries to "continue" still no response and now out of my weekly usage on the Pro plan.

■ {
"error": {
"message": "The encrypted content gAAA...Vg8W could not be verified.",
"type": "invalid_request_error",
"param": null,
"code": "invalid_encrypted_content"
}
}

Same error. I'm using Codex CLI on PowerShell; in one instance it generates the error mentioned in the issue: "The encrypted content gAAA...Lzo= could not be verified." In another instance, it tries to reconnect 5 times ("Reconnecting... 1/5") and then shows: "We're currently experiencing high demand, which may cause temporary errors."

Same error resuming previous chat. A new chat works.

Same error, filed a dupe by accident: https://github.com/openai/codex/issues/8132

Good luck!

Same issue for me, not resuming just starting a new one

``■ {
"error": {
"message": "The encrypted content gAAA...xPbT could not be verified.",
"type": "invalid_request_error",
"param": null,
"code": "invalid_encrypted_content"
}
}``

• Installed Codex CLI via npm and verified version
• sudo npm install -g @openai/codex
• codex --version
• which codex
• Authenticated successfully
• codex auth login
• Verified token: ls ~/.codex (shows auth.json)
• Reproduced invalid_encrypted_content error consistently (including on inline-pasted text)
• Fully quit Codex and killed any running processes

• pkill -f codex || true
• Cleared Codex local state (including full wipe)

• rm -rf ~/.codex
• (also tried) rm -rf ~/.codex/sessions
• (also tried) rm -rf ~/.codex/cache
• Fully uninstalled and reinstalled Codex

• sudo npm uninstall -g @openai/codex
• sudo npm install -g @openai/codex
• Downgraded Codex versions

• sudo npm install -g @openai/codex@0.72.0

› can you read any files//

• Yes — I can read files in this repo from /mnt/c/Users/Jarad/Documents/PyCharm/myproj (and elsewhere, since this run is danger-full-access).

  Tell me which file(s) you want (e.g. app/manage.py, AGENTS.md, app/myproj/settings.py), or what you’re looking for, and I’ll open them.


› Read TODO.md


■ {
  "error": {
    "message": "The encrypted content gAAA...gD09 could not be verified.",
    "type": "invalid_request_error",
    "param": null,
    "code": "invalid_encrypted_content"
  }
}


› lol

Is there a fix or it yet?
Facing the same issue

■ {
  "error": {
    "message": "The encrypted content gAAA...OQ==
could not be verified.",
    "type": "invalid_request_error",
    "param": null,
    "code": "invalid_encrypted_content"
  }
}

Thank you!!

downgrading worked for me so far

This is a server-side issue. Upgrading or downgrading the CLI will not fix this. If that seems to fix it, it's by chance only.

We understand the issue, and we're working on testing a fix and deploying it as quickly as possible.

The fix is rolling out across our server clusters. Service should be restored for everyone shortly. Continuing to monitor the situation closely.

I can confirm it works for me now

I'm still seeing errors

<img width="713" height="161" alt="Image" src="https://github.com/user-attachments/assets/1b780d53-5cef-4454-837d-fafdde33b44d" />

<img width="625" height="468" alt="Image" src="https://github.com/user-attachments/assets/32cca902-6e8a-4dd0-91d2-e0a53537d2ec" />

OpenAI Status

Temporary fix: just say "continue" and have codex try again (keep going).

Update: People report this not working for them, but it did work in my case, I guess the requests hit services that didn't have this bug. Requests are usually routed to different machines based on different factors like geography, so I think it's just a matter of waiting for the update to be fully propagated now.

Typing "continue" fixes nothing. It just repeats the same error, over and over:

{
"error": {
"message": "The encrypted content gAAA...Pg== could not be verified.",
"type": "invalid_request_error",
"param": null,
"code": "invalid_encrypted_content"
}
}

Continue doesn't work for me either. Starting a new session in the same folder works, but then I lose context.

Typing “continue” should just make the service run the check once again to see if it’s back to normal.
It looks like we can monitor the recovery status here: https://status.openai.com/

Continue is a hoax

Continue doesn't work for me

continue doesn't work

They mentioned it's a server side problem and that they already deployed a solution to their clusters, I can imagine that different requests land on different machines, so that's why continue worked for me, it tried again and it worked. I'm also guessing requests are routes geographically among different clusters, and I'm also guessing that updating all those clusters does take some time. I didn't mean to provide a bad solution, it's just that it really did work for me.

Note that this error was always only showing after a few minutes running in my case and not always immediately.

continue seems to kind of work... until it doesn't. I am tempted to just add an absurd number of retries instead of the default of 5. it seems to work every once in a while.

still an issue:
<img width="688" height="167" alt="Image" src="https://github.com/user-attachments/assets/148e8d5f-bb15-4e93-87b3-
22f2be2cc87f" />

Good thing I still know how to code, otherwise I would be stuck waiting for this vaporware-as-a-service thing to work.

Using: v0.73.0

Change the default retries to something like request_max_retries = 999 and code by hand until it wants to do what you ordered.

Still an issue. On OpenAI status page it only mentions ChatGPT but I also have this behavior using Codex GitHub action so using an API key.
If it can help I submited thread ID 019b2928-98de-7d63-9abf-4ceba53f3745 using CLI where I used the same prompt which fails with the same error using Codex action.

Continue works, and I tried 3 times then it works.

I have luck 1 out of 10 times , so let's hope it will resolve soon

I can confirm this has just happened to me again a couple of minutes ago from a new prompt that worked for a while, so not fixed yet

Same problem her in the Codex extension in VSCode:

• Codex extension version 0.4.51
• macOS Darwin 24.6.0 (arm64)

As others mentioned, "continue" works until it does not.

Whatever new fix was implemented 20minutes ago, just broke everything again.

Can you fix your flagship product or are you charging us for beta testing it for you?

Codex quality is going down the drain and now this is happening?

also using extension version 0.4.51
macos sonoma 14.6.1 (23G93)

might be a con incident but i used vpn and now it has started working again

The fix has been deployed to most of our server clusters, but there are still issues remaining. It should be working for most codex users at this point. I realize that's little consolation for those of you who are still affected. The team is continuing to work on getting the service to a healthy state across the board.

The fix has been deployed to most of our server clusters, but there are still issues remaining. It should be working for most codex users at this point. I realize that's little consolation for those of you who are still affected. The team is continuing to work on getting the service to a healthy state across the board.

Are we going to reimbursed for the tokens and progress lost due to this?

The fix is deployed across all clusters. All new conversations should now succeed without hitting the problem. You may still experience the error if you resume conversations that were started prior to the incident. If you still see the problem with a new conversation, let us know (use /feedback and post the thread ID).

> The fix has been deployed to most of our server clusters, but there are still issues remaining. It should be working for most codex users at this point. I realize that's little consolation for those of you who are still affected. The team is continuing to work on getting the service to a healthy state across the board. Are we going to reimbursed for the tokens and progress lost due to this?

Seconding this! Had multiple sessions get context-corrupted due to the issue

A mi todavia me sale el error, cierro sesion o que hago ?
{ "error": { "message": "The encrypted content gAAA...qgo= could not be verified.", "type": "invalid_request_error", "param": null, "code": "invalid_encrypted_content" } }, ya mIre perdio el contexto

> > The fix has been deployed to most of our server clusters, but there are still issues remaining. It should be working for most codex users at this point. I realize that's little consolation for those of you who are still affected. The team is continuing to work on getting the service to a healthy state across the board. > > > Are we going to reimbursed for the tokens and progress lost due to this? Seconding this! Had multiple sessions get context-corrupted due to the issue

I also burned tokens retrying and lost context. They should reimburse tokens/credits consumed due to retries during the incident

New conversations are working fine but I've lost the context-progress in earlier conversations which is a shame. Are burnt tokens going to be added back to our usage? That would be very helpful.

We've reset usage limits for subscribers. To those who were affected by this incident, we apologize for the disruption.

We've reset usage limits for subscribers. To those who were affected by this incident, we apologize for the disruption.

I need to request the reset of usage limits, or will be an automatic action?

Issue still exist on latest release version.

How can I debug to fix resolve it?

OpenAI Codex (v0.76.0)
model: gpt-5.2 high

{"error":{"message":"The encrypted content gAAA...0NgW could not be verified.","localized_message":"Unknown
error","type":"invalid_request_error","param":"","code":"invalid_encrypted_content"}}