## Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

## What happened?

Every non-admin user logged into the LiteLLM dashboard generates a continuous stream of ERROR-level log entries with full stacktraces — several per page load/refresh. The dashboard unconditionally calls admin-only endpoints on every page render, regardless of the logged-in user's role.

**Affected routes (called by the UI for all users):** - `GET /get/mcp_semantic_filter_settings` - `GET /tag/list` - `GET /user/daily/activity/aggregated` - `GET /policies/list` - `GET /config/list` - `GET /credentials` - `GET /get/config/callbacks` - `GET /prompts/list` - `GET /in_product_nudges`

Each of these correctly returns 403 for `internal_user` role — authorization works. The problem is that `auth_exception_handler.py` logs every 403 via `verbose_proxy_logger.exception()` at ERROR level with a full stacktrace, indistinguishable from real application errors.

## Steps to Reproduce

1. Create a user with `internal_user` role 2. Log into the LiteLLM dashboard with that user 3. Navigate to any dashboard page 4. Observe logs: each page load generates ~10–15 ERR…