How to Disable Confirmation Requirement for External Service Calls in Custom GPT with Actions?

GPT builders
, , ,
1

Hi everyone,

I’m using actions in my custom GPT model through ChatGPT, and I encounter a bit of an issue that I’m hoping to resolve. Every time I attempt to send data to an external service using an action, I’m prompted to confirm this manually.

I’ve noticed that in the “GPT’s privacy settings” for the custom chat, there’s an option to select “Always Allow,” which seems like it should bypass the repeated confirmation prompts. However, despite setting it to “Always Allow,” I’m still getting the manual confirmation request each time I trigger an action to send data to an external service.

I’ve attached some screenshots illustrating the settings:

  1. Confirmation prompt I’m trying to avoid.
  2. “Always Allow” option in privacy settings.
  3. Privacy settings access through the chat menu.

My goal is to configure it such that once I allow actions in a specific chat, I no longer need to manually confirm each subsequent request. How can I achieve this for a particular custom chat, so it automatically allows the external calls without prompting every time?

Any help or suggestions would be greatly appreciated!

Thanks in advance!

image
image698×236 49.5 KB

image
image1052×474 56.8 KB

image
image743×324 58.7 KB

1 Like
2

+1. Seems super limiting - it stops us using voice chat for trusted customGPTs with actions.

2 Likes
3

Yeah, its super limiting in user experience I agree. Leaving this reminder here to be noticed :wink:

1 Like
4

It might be better to prompt once for all at the start of the conversation. It is indeed totally breaking the voice experience. When prompted once, you must start a new chat to keep going which is of course not the point!

1 Like
5

UPDATE:

  • Found : “If you’re using an openApiSchema then you would set x-openai-isConsequential to false”

worked :slight_smile:

PS. Would post the link but even internally, they dont let us do that :frowning:

1 Like
6

Thanks! Great update (y)

This is a note from docs:

  • If the x-openai-isConsequential field is true, ChatGPT treats the operation as “must always prompt the user for confirmation before running” and don’t show an “always allow” button (both are features of GPTs designed to give builders and users more control over actions).
  • If the x-openai-isConsequential field is false, ChatGPT shows the “always allow button”.
  • If the field isn’t present, ChatGPT defaults all GET operations to false and all other operations to true

I’ve checked the POST request with x-openai-isConsequential: false - it works!