Bicep deployments blocked by Cognitive Services fraud prevention error 715-123420

Resources below were already deployed but now I am getting the following error when reapplying the bicep template for making changes to other resources. My tickets to change the quota keep getting rejected as I don't have a technical support plan. This is not a technical issue and the block needs to be lifted by Microsoft.

ERROR: { 

  "error": { "code": "InvalidTemplateDeployment", 

"message": "The template deployment 'templateName' is not valid according to the validation procedure. The following resource provider(s) - 'Microsoft.CognitiveServices/accounts (2026-07-15-preview)' reported preflight validation errors. Tracking id is 'xxxxxx'. See inner errors for details.", 

"details": [ 

  { 

    "code": "715-123420", 

    "message": "Our system has detected this request as unusual activity for your account. If you are confident this is in error, please contact support." 

  } 

] 
  }, 

  "properties": {} 

} 
```except from main.bicep:

  

```bicep
// Document Intelligence (Form Recognizer)
resource documentIntelligence 'Microsoft.CognitiveServices/accounts@2026-07-15-preview' = {
  name: 'di-${projectEnvLocName}'
  location: rgLocation
  kind: 'FormRecognizer'
  tags: {
version: deployment().properties.template.contentVersion
  }
  sku: {
name: (envName == 'prod' || envName == 'uat') ? 'S0' : 'F0'
  }
  properties: {
customSubDomainName: 'di-${projectEnvLocName}'
publicNetworkAccess: 'Enabled'
  }
}
// Azure OpenAI
// NOTE: the model must be available in rgLocation. If it isn't, give this account its own location param.
resource openAi 'Microsoft.CognitiveServices/accounts@2026-07-15-preview' = {
  name: 'oai-${projectEnvLocName}'
  location: rgLocation
  kind: 'OpenAI'
  tags: {
version: deployment().properties.template.contentVersion
  }
  sku: {
name: 'S0'
  }
  properties: {
customSubDomainName: 'oai-${projectEnvLocName}'
publicNetworkAccess: 'Enabled'
  }
}
// Chat model deployment. Its name is what AiGateway:Ai:AzureOpenAi:Model points at.
// Shared with apps.bicep via ./variables/llm.json so the name/version live in one place.
// Tune capacity (TPM, in thousands) per region availability and quota.
var llm = loadJsonContent('./variables/llm.json')
{
  "modelDeploymentName": "gpt-5.4-nano",
  "modelVersion": "2026-03-17"
}
resource openAiChatDeployment 'Microsoft.CognitiveServices/accounts/deployments@2026-07-15-preview' = {
  parent: openAi
  name: llm.modelDeploymentName
  sku: {
name: 'GlobalStandard'
capacity: 30
  }
  properties: {
model: {
  format: 'OpenAI'
  name: llm.modelDeploymentName
  version: llm.modelVersion
}
  }
}