Ask HN: Why does OpenAI require an ID to use their image API?
4 comments
It's weird because I can do this and I'm sure I never sent them any verification. Because I'm always incredibly hesitant to do so, it would have been a dealbreaker. Our national IDs contain a lot of sensitive info. The government even advises us to black those out. But then most companies refuse it.
I do have credits paid up though and I've had an account since the early days of DALL-E. Maybe having a payment method serves as verification? Or it only applies to recently created accounts?
I do have credits paid up though and I've had an account since the early days of DALL-E. Maybe having a payment method serves as verification? Or it only applies to recently created accounts?
I think part of the reason is that they want to be able to track the images back to you. Some person in the company is probably worried that a user will generate child porn like images and distribute them on the net. When the cops eventually come calling, they want to point to the user. This probably means there is a user-id embedded somewhere in the image.
If you put the feature behind a paywall, technically you can track requests using the payment card ?
In that case, what would be the added benefit of using an government issued ID ? really curious.
In that case, what would be the added benefit of using an government issued ID ? really curious.
OpenAI has always been somewhat controlling of their output. They used to require that anything that uses their API go through a verification process to prevent it from being used for fake reviews, astroturfing, etc. This was the main criticism of OpenAI 3 or so years ago. But with all the other alternatives, it hasn't stopped people.
I suppose while they have the best image generator of the quarter, they'll be more restrictive with it.
I suppose while they have the best image generator of the quarter, they'll be more restrictive with it.
I was planning to build a couple of web applications on top of the API but this is really off putting, especially given how prevalent data breaches have become and the simple fact that there’s no need for it other than their pathetic attempts to look like they care about AI safety.