We have an approved ChatGPT app that’s getting its tool calls blocked for “safety” reasons after publishing it for the general public. It seems that this is happening because the tool accepts URLs as inputs (?!) but I’m just guessing and not sure.
We have never faced this issue during testing, neither in Developer Mode, nor when the app was published in the internal ChatGPT workspace of our company. The app review also went smoothly and our app was approved by OpenAI.
- Can you confirm that tool inputs containing URLs is the reason the tool calls are getting blocked?
- If so, what’s the guidance here? Redesign the tool’s input schema to avoid URLs?
Thank you