BadrequestError

Question

BadrequestError

Jun 0

Created a chatbot to query company document, it works fine until the query contains some molecule ID, and system shows the error below:

how to solve this issue? the document is medicine development purpose and did not contain any violence content.

I am using GPT-4 with openAI_API_version:2023-03-15-preview

BadRequestError: Error code: 400 - {'error': {'message': "The response was filtered due to the prompt triggering Azure OpenAI's content management policy. Please modify your prompt and retry. To learn more about our content filtering policies please read our documentation: https://go.microsoft.com/fwlink/?linkid=2198766", 'type': None, 'param': 'prompt', 'code': 'content_filter', 'status': 400, 'innererror': {'code': 'ResponsibleAIPolicyViolation', 'content_filter_result': {'hate': {'filtered': False, 'severity': 'safe'}, 'self_harm': {'filtered': False, 'severity': 'safe'}, 'sexual': {'filtered': False, 'severity': 'safe'}, 'violence': {'filtered': True, 'severity': 'medium'}}}}}

During handling of the above exception, another exception occurred:

navba-MSFT 27,625 Reputation points Microsoft Employee Moderator

2024-03-24T05:02:58.7833333+00:00

@Jun Just following up to check if my suggestion helped. Please let me know if you have any further queries. I would be happy to help.

2 answers

Your answer

navba-MSFT 27,625 Reputation points Microsoft Employee Moderator

2024-03-24T05:02:58.7833333+00:00

@Jun Just following up to check if my suggestion helped. Please let me know if you have any further queries. I would be happy to help.

Answer 1

@Jun Welcome to Microsoft Q&A Forum, Thank you for posting your query here!

You can follow the below action plan to address this:

Plan 1:
Try to modify the input prompt, such as adding a description to tell the model that the following is not true, or removing the centered sensitive word through preprocessing. Please note that this method may not be applicable in all scenarios.

Plan 2:

If the above doesn't help. You need to request an opt-out of the content moderation mechanism. If your application scenario will inevitably encounter sensitive words, it is recommended that you apply directly and withdraw from the review mechanism, this method will be more stable, and we do not need to modify the prompt due to changes in user input.

Please note that this application is manually reviewed by the Microsoft CSGATE team, so please be sure to fill it out carefully.

You need to request to modify content filtering and/or abuse monitoring by submitting the form at https://aka.ms/oai/modifiedaccess

After you submit this form, we review it and respond in approximately 10 business days. Some reviews may take longer than 10 business days for a final decision.

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.

Answer 2

Azar 31,720 MVP Volunteer Moderator

Hey there Jun

Thanks for posting the question in QandA platform\

So based on the error message you gave, I see that the content filtering policy is blocking content related to "violence." The error message specifically states that the content related to violence has been filtered with a severity level of "medium."

I understand you are using good data and prompt but still that contains elements or references related to violence, which is being flagged by the content filtering policy.

I suggest you to review and modify the query prompt to remove or adjust any language or references that could be interpreted as promoting or containing violent content.

Ill drop the link to the azure ai content filtering document. go through them for more details.

https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/content-filter?tabs=warning%2Cpython-new

If this helps kindly accept the answer thanks much.,

Jun 0 Reputation points

2024-03-21T19:29:49.9466667+00:00

Thank you for your kind information. The query is standard simple question "what is "molecule ID"?" and content includes some pathology report how this molecule impact the human tissues. This application is used for pharmacy study, and should not be defined as the violence content. Any chance to configure the GPT content policy?
Azar 31,720 Reputation points MVP Volunteer Moderator

2024-03-21T19:34:37.62+00:00

Hi again Jun

I understand but at this point of time, I guess we don't have much information about this and cant bypass this content filtering, ill keep an note on this and if anything comes up ill let you know, meanwhile you can try with diff data or diff prompts.

Kindly accept the response if it helps thanks again, have a good day.

Jun 0

one more thing, because this BadRequestError 400 happened, it stops the app without returning any information. I tried to use:def agent_node(state, agent, name):

try:

    result = agent.invoke(state)

    return {"messages": [AIMessage(content=result["output"], name=name)]}

except BadRequestError:

    abort(400)

    result = "Content policy triggered, please rephase your question."

    return {"messages": [AIMessage(content=result["output"], name=name)]}

To return the message to users instead of nothing, but it not working, still showing

BadRequestError

and

NameError: name 'BadRequestError' is not defined

any advice on this? thanks

Azar 31,720 MVP Volunteer Moderator

Not sure but try importing this from azure.core.exceptions import BadRequestError

like this

from azure.core.exceptions import BadRequestError
def agent_node(state, agent, name):
    try:
        result = agent.invoke(state)
        return {"messages": [AIMessage(content=result["output"], name=name)]}
    except BadRequestError:
        result = "Content policy triggered, please rephrase your question."
        return {"messages": [AIMessage(content=result, name=name)]}

Jun 0 Reputation points

2024-03-21T20:49:26.3766667+00:00

many thanks

but the exceptions didn't include BadRequestError....

double check the lib, which includes the below:

"AzureError",

"ServiceRequestError",

"ServiceResponseError",

"HttpResponseError",

"DecodeError",

"ResourceExistsError",

"ResourceNotFoundError",

"ClientAuthenticationError",

"ResourceModifiedError",

"ResourceNotModifiedError",

"TooManyRedirectsError",

"ODataV4Format",

"ODataV4Error",

"StreamConsumedError",

"StreamClosedError",

"ResponseNotReadError",

"SerializationError",

"DeserializationError",

]
Jun 0 Reputation points

2024-03-21T20:53:40.8933333+00:00

redundant content
Azar 31,720 Reputation points MVP Volunteer Moderator

2024-03-22T15:34:17.0533333+00:00

If the above response helps kindly accept the answer thanks much.

Share via

BadrequestError

2 answers

Your answer