Without warning, OpenAI suddenly implemented a large-scale account ban.

Woke up to find multiple GPT accounts have been banned!

OpenAI suddenly launched a mass ban spree, completely deleting and deactivating accounts directly without any email notification.

There is only one error message on the screen:「Your account has been deleted or deactivated (You do not have an account because it has been deleted or deactivated)」.

The worst-hit regions are the Philippine region, Turkish region, low-cost Team accounts, and accounts using reverse proxy relays. Even many normal long-term accounts that have been renewed for four or five months have been affected.

Shortly after Cat Wu moved from Anthropic to OpenAI, ChatGPT immediately launched a large-scale account ban wave.

Netizens' inner thoughts: Please, let us go!

All accounts that have been used for years are gone

Different from Anthropic, this massive purge by OpenAI covers free API accounts, team shared accounts, and $200/month Pro subscriptions.

A large number of programmers who were halfway through writing code immediately exploded, flocking to the X comment section of Tibo, former head of OpenAI Codex, to post messages for rights protection.

One developer complained helplessly that years of conversation records and personal memories he had painstakingly accumulated were all locked in the cloud and could not be retrieved.

A Pro account that had been used for five full years was deactivated, with no warning, no email, and no reason given.

And appeals got no results.

At present, Pro 20x, Pro, some Team, Plus accounts, and even student verified accounts have been banned,

Even overseas, accounts used normally may also get banned.

Some people speculate that this is OpenAI's move to prevent users from taking advantage of loopholes to get free benefits, and it has started to limit the maximum number of accounts per device.

Some people suspect that it is a payment problem —

OpenAI status page shows that some US commercial Mastercards have a high error rate.

Under investigation, the first six digits of the affected card numbers are 544015, 546400, 543502, 543799, 512998, and some cards starting with 513989 are also affected.

However, OpenAI defined the incident as a payment channel problem and did not mention account bans.

Some netizens suggested that you can appeal, you do not have to send emails from your registered email, just use the official OpenAI appeal form directly. You need to clearly write the information of the banned account, and the Request ID is in the red box in Figure 1).

Previously, he successfully appealed to get back his student verified Plus account!

The same story is happening all over again

This move by OpenAI is by no means sudden.

Previously, they launched reverse proxy detection, and merged ChatGPT and Codex into one Token pool.

Five days ago, they quietly locked the upper limit of paid accounts for a single device.

It is getting harder and harder to open low-cost accounts to exploit loopholes for free usage. Until today, the ban hammer has fallen completely!

Specifically, on November 2, OpenAI blocked "stacked accounts" at the hardware device and bank card underlying levels.

An official document titled "Understanding ChatGPT Payment Card and Device Limits" launched 5 days ago brought all tiers including Go, Plus, Pro 100, Pro 200, and Pro 500 under strict quota control.

As long as one payment card or one device is associated with too many paid subscriptions, the prompt Too many subscriptions associated with this card or device will pop up during checkout, and you cannot complete the purchase.

In addition, the maximum number of paid accounts that can be used on a single device has also been set.

Once the Device limit reached prompt is triggered, even if you immediately log out of the old account, or even pay extra to buy Credits, you cannot free up the quota on this device.

This combined set of measures prevents users from opening multiple cheap accounts to stack Codex quotas.

The second step is the large-scale purge of shared accounts today.

Users who tried to integrate multiple accounts through APIs have all been banned in one go today.

And the most critical third move is to restrict computing power exports, scheduled for late November to early December.

Starting from November 20, OpenAI's new "Terms of Service" will take effect, mandating that all third-party API gateways and model routing intermediaries must obtain official written authorization, banning unofficial relay stations from both legal and interface levels.

In addition, starting from December, ChatGPT and Codex for Plus, Pro and Business plans will share the same quota, and will be fully calculated based on Token consumption!

Once the Token pool is exhausted, you cannot rely on switching to alternate accounts to continue using the service, you can only upgrade your plan honestly or purchase official credits.

What users can do is to export cloud sessions regularly and synchronize Codex projects to local storage. Top-up through third parties, shared accounts, and unofficial relays are all completely blocked.

What we experienced at Anthropic before is happening again today.

The wild era of free exploitation of AI computing power is over.

References:

https://help.openai.com/en/articles/20001625-understanding-chatgpt-and-codex-usage-limits

https://x.com/gudabing/status/2118960197800513858

https://x.com/AoxClaw/status/2118943001540174090

https://x.com/divid_lu49394/status/2118130977553969289

This article is from WeChat Official Account "Xinzhiyuan" (ID: AI_era), written by ASI Revelation, authorized for release by 36Kr.