> ## Documentation Index
> Fetch the complete documentation index at: https://docs.abliteration.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Policies in API requests

> Use published policies with the standard Chat Completions, Responses, and Messages APIs.

Policies run on the standard inference endpoints:

| API | Endpoint |
| - | - |
| Chat Completions | `POST /v1/chat/completions` |
| Responses | `POST /v1/responses` |
| Messages | `POST /v1/messages` |

Published policies scoped to your organization or project apply automatically. You do not need to change the endpoint, API key, or request body to use them.

To add a published policy configured as **Only when selected**, include its ID in the top-level `policy_ids` array. The field is available on all three inference endpoints:

```json theme={"system"}
{
  "model": "abliterated-model",
  "messages": [{"role": "user", "content": "Hello"}],
  "policy_ids": ["pol_YOUR_POLICY_ID"]
}
```

This example is a Chat Completions body. For Responses or Messages, keep their normal request shape and add `policy_ids` at the top level. IDs must refer to published policies available to the caller's organization and project. The field does not replace policies that already apply automatically.

If a policy blocks or fails, the API returns its normal endpoint-specific [error format](/api/errors#policy-errors). A response policy can delay output, including streamed output, until evaluation finishes. See [Use policies](/policy-gateway/integration).