GPT-5 streaming requires submission of biometric data

33 points by binarymax 2 days ago

OpenAI is requiring people to submit their biometric information to Persona, if you want to use GPT-5 in streaming mode through the API.

This is the error you will receive:

  error: {
    message: 'Your organization must be verified to stream this model. Please go to: https://platform.openai.com/settings/organization/general and click on Verify Organization. If you just verified, it can take up to 15 minutes for access to propagate.',
    type: 'invalid_request_error',
    param: 'stream',
    code: 'unsupported_value'
  }
Clicking through and pressing verify you are challenged to submit your ID and with this message:

> By filling the checkbox below, you consent to Persona, OpenAI’s vendor, collecting, using, and utilizing its service providers to process your biometric information to verify your identity, identify fraud, and conduct quality assurance for Persona’s platform in accordance with its Privacy Policy and OpenAI’s privacy policy. Your biometric information will be stored for no more than 1 year.

esafak 2 days ago

Are you sure it is only for streaming? No GPT5 for me then. Can you imagine if companies started mandating employees get fingerprinted just to use ChatGPT, whose use is also mandated? No, thanks.

  • binarymax 2 days ago

    Yes. I found this out because I have a small tool I maintain for OpenAI API integration - and my integration test suite triggered this for the streaming case. The non-streaming cases didn't trigger the requirement.

    FWIW I am definitely not opting in to this overreach and will use a different provider if necessary. I currently use GPT-4o for some streaming apps I have, but will migrate away from this requirement.

  • muzani 2 days ago

    There's some other bits like the image endpoint as well.

    • XenophileJKO a day ago

      My impression is that they have been activating this for endpoints that expose more risk for misuse. They are trying to keep bad actors off. Streaming likely is protected because response classification often lags behind the stream.

tempodox 2 days ago

Looks like the data grabbers found themselves a means to extort the juiciest bits from their victims. I'm sure the majority have “nothing to hide”.

mikeodds 2 days ago

worldcoin OpenAI flywheel