I'm getting this error code:
{"operationStatus":"Error","error":{"type":"Error","code":"InputContentFiltered","message":"Prompt was filtered. [105]","properties":{"BackendErrorCode":"InputContentFiltered","DependencyHttpStatusCode":"400"}},"predictionId":null}
I think it is encountering what it believes is unsafe content, but there most definitely is not. How can get confirmation that this is the case, what content it believes is unsafe, etc.?
Sorry to read that you are facing this issue, would you please share the prompt with me in a private message?
Thanks
Regards
Philippe
This IMO / IME - I get the same thing and my prompt is "Tell me about Microsoft AI" ... SERIOUSLY!!! you're filtering this???
The "Create text with GPT using a prompt" action in power automate "as is", is worthless.
Even adding an extra action "Start and wait for an approval of text" after it it still does not work and fails.
This is why I use OpenAI directly, its WAY cheaper or I use private LLM's like llama3. I need to get work done, not deal with some overly hyper sensored AI that fails all input. This is an extreme over configuration IMO.
I'm experiencing the same thing. What's the work around?
I always receive the same error and it's really frustrating. How to resolve this?
Yes, we have added very restritive rules to protect our customers and Microsoft.
It is possible that in some cases you may encounter false positive (safe content was filtered) but we optimized to avoid false negative (Unsafe content wasn't filtered)