azure-openai-inference-response-content
Model responses can reproduce submitted sensitive content or generated credentials/code.
Where it sits
| location | Azure OpenAI response.{choices,output,data,usage} |
|---|
| location kind | output_value |
|---|
| data kinds | credential password api_key access_key secret_key oauth_token connection_string source_code_secret sensitive_data pii customer_data |
|---|
| emits edge | ContainsSecret |
|---|
| service | Azure OpenAI / AI Services (azure:openai) |
|---|
Collection recipe
| access mode | data_plane |
|---|
| operation | POST deployment inference operation |
|---|
| response path | {choices,output,data} |
|---|
| encoding | json |
|---|
| params | {"resource_id": "\u003cazure-resource-id\u003e"} |
|---|
Required permissions
Microsoft.CognitiveServices/accounts/OpenAI/deployments/chat/completions/action
References