BerriAI/litellm · error · Exception

Trying to use ENTERPRISE BannedKeywordYou must be a LiteLLM

Error message

Trying to use ENTERPRISE BannedKeywordYou must be a LiteLLM Enterprise user to use this feature. If you have a license please set `LITELLM_LICENSE` in your env. Get a 7 day trial key here: https://www.litellm.ai/enterprise#trial. 
Pricing: https://www.litellm.ai/#pricing

What it means

The 'banned_keywords' callback loaded its enterprise module, but the license gate failed: premium_user is not True at config load time. The message combines 'Trying to use ENTERPRISE BannedKeyword' with the standard not_premium_user text that points to LITELLM_LICENSE and the trial link. Startup aborts at this raise.

Source

Thrown at litellm/proxy/common_utils/callback_utils.py:305

                    )

                if premium_user is not True:
                    raise Exception("Trying to use ENTERPRISE BlockedUser" + CommonProxyErrors.not_premium_user.value)

                blocked_user_list = _ENTERPRISE_BlockedUserList(prisma_client=prisma_client)
                imported_list.append(blocked_user_list)
            elif isinstance(callback, str) and callback == "banned_keywords":
                try:
                    from enterprise.enterprise_hooks.banned_keywords import (
                        _ENTERPRISE_BannedKeywords,
                    )
                except ImportError:
                    raise Exception(
                        "Trying to use Banned Keywords" + CommonProxyErrors.missing_enterprise_package_docker.value
                    )

                if premium_user is not True:
                    raise Exception("Trying to use ENTERPRISE BannedKeyword" + CommonProxyErrors.not_premium_user.value)

                banned_keywords_obj = _ENTERPRISE_BannedKeywords()
                imported_list.append(banned_keywords_obj)
            elif isinstance(callback, str) and callback == "detect_prompt_injection":
                from litellm.proxy.hooks.prompt_injection_detection import (
                    _OPTIONAL_PromptInjectionDetection,
                )

                prompt_injection_params = None
                if "prompt_injection_params" in litellm_settings:
                    prompt_injection_params_in_config = litellm_settings["prompt_injection_params"]
                    prompt_injection_params = LiteLLMPromptInjectionParams(**prompt_injection_params_in_config)

                prompt_injection_detection_obj = _OPTIONAL_PromptInjectionDetection(
                    prompt_injection_params=prompt_injection_params,
                )
                imported_list.append(prompt_injection_detection_obj)
            elif isinstance(callback, str) and callback == "batch_redis_requests":

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Export a valid LITELLM_LICENSE for the proxy process and restart.
2. Or drop 'banned_keywords' from the callbacks list.
3. In Kubernetes/compose, verify the env var is in the pod spec and not only in the local shell.

Example fix

# before (compose, no license)
environment:
  - LITELLM_MASTER_KEY=sk-...

# after
environment:
  - LITELLM_MASTER_KEY=sk-...
  - LITELLM_LICENSE=sk-...

import os

def validate_config(cfg: dict) -> None:
    cbs = (cfg.get("litellm_settings") or {}).get("callbacks") or []
    if "banned_keywords" in cbs and not os.environ.get("LITELLM_LICENSE"):
        raise SystemExit("banned_keywords needs LITELLM_LICENSE")

try:
    start_proxy(cfg)
except Exception as e:
    if "ENTERPRISE BannedKeyword" in str(e):
        raise SystemExit("license missing: export LITELLM_LICENSE or drop banned_keywords")
    raise

When it happens

Trigger: Proxy runs on the enterprise Docker image with callbacks: ['banned_keywords'] while LITELLM_LICENSE is absent, expired, or invalid.

Common situations: The image is right but the license env var never reached the container. The license expired after a billing change. Staging works, production fails because only staging got the env var.

Related errors

• Trying to use Google Text ModerationYou must be a LiteLLM En
• Trying to use Llm GuardYou must be a LiteLLM Enterprise user
• Trying to use ENTERPRISE BlockedUserYou must be a LiteLLM En
• Trying to use Banned KeywordsThis uses the enterprise folder
• You must be a LiteLLM Enterprise user to use this feature. I

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/0767363883eb0069. Report an issue: GitHub.