Spontaneous gpt-4 migration and pricing started to increase rapidly

I specifically selectively use gpt-3.5-turbo in my current applications. But as of yesterday, requests started to be sent automatically with gpt-4 and this caused my charges to increase rapidly. I have passed more than my total billing of 2 months in 12 hours. I can’t reach anyone and I can’t stop it. Because I didn’t want to switch to gpt-4! Although it is not my fault, I will now have to stop the service I provide to my users because the fee is increasing at an incredible rate. Please contact me if someone can help with this.

The price continues to rise rapidly. Why isn’t anyone helping?

I deleted my APIKEY but 7,000 tokens are still going in one request over gpt4 and the fee continues to increase. What is this nonsense? It’s not possible for anyone to send requests with my API_KEY. I just created it 5 minutes ago and didn’t add it to any of my apps. OpenAI doesn’t have enough staff? Why is no one responding to emails and this topic?

I am sorry to hear that. I suggest you limit the losses and cut the cord until things get sorted out.

Yesterday I received response from customer service within a few hours but others have reported weeks of silence. You can’t bank on any reply, that’s why I suggest to proceed by turning it off.

I mean who knows? It could still get worse…

Unfortunately, by deleting my API_KEY, I cut support for my most used and growing application. I am really depressed. OpenAI’s site created for API use does not provide enough information, using which API was the request made? Was this request fired from my current app or from an external app using my key? There are no answers to such questions. It is also impossible to reach them. However, we pay for the service.

My question would be, how is your application suddenly using GPT-4 now? It has to be specifically requested as the model to use in each request. It’s not something that just changes unless it’s written to do that.

1 Like

Yes, I chose gpt-3.5-turbo as the model for each request, but some requests go with gpt-4. I think my key was stolen and someone is sending high token requests via gpt-4. I had to delete the key and now the requests are gone. Although I have never shown the key anywhere, I have no idea how it was used, be careful.

Is it online application, or is it executed by the client? How do you use the key, where was it stored?

1 Like

Yes, my app in the Play Store. I was storing the key in the .env file in my app and keeping the project private on github. But I really don’t know how they got my key. At first, I thought that the requests in my application were doing this, but I already made the necessary settings in my application so that there was not such a high amount of tokens. So I think the request was made from a different place.

If the key was stored inside the app it’s easy for bad actors to extract your key and use it any way they’d like. Your key should instead be securely stored on your own backend server that you call from the app.

3 Likes

Exactly! And now you can fix that and restart with next update!

Good luck, I guess.

1 Like

I didn’t know it would be this easy. It was an expensive experience. I will do as you say.

1 Like

There are hundreds (if not thousands?) keys available in the internet, extracted from applications like yours. No API keys, ever, should be deployed to customer workstation, in any form.
That’s true for all APIs, including Azure, AWS and so on. You know what? Have a chat with ChatGPT about it. It will tell you about why it’s a bad idea and what you can do about it.

P.S. You can do some google search for your key. Just for the fun of it.

1 Like

I am having the same issue. Even after stopping all our app use (which was only being used by our test team anyways), we are still getting billed in $100s of $$$

We literally can chart a MASSIVE spike in past 4 days which is abnormal. Our app is not even public and keys are stored in backend. No word at all from OpenAI

Please bare with it, the support team is under immense load from all of the new API users, your queries will be responded to in due course.

I am again facing the same issue. It happened last month and luckily got a refund, but again GPT-4 bills started out of nowhere. OpenAI has a major bug. We are using industry standard for saving and protecting our keys so there is no way, there is any leak.
If there is any OpenAI rep here, you need to investigate.