1. Private by default
LizzFlix does not use your prompts, uploads, or outputs to train LizzFlix-owned models.
Content is still processed by third-party generation providers when you intentionally choose a model and start a job. They process it under their own terms, and some of them may use it to improve their models, as described below.
2. Provider training policies
- Runway: its terms of use allow Runway to use inputs and outputs to train and improve its AI models.
- fal: does not train on your content, except for models fal lists as "Pending Enterprise Ready". The third-party models that fal runs receive the content you send to them.
- Replicate: its terms give Replicate a licence to use content to provide outputs and to train "Customer Derivative Models".
- ElevenLabs: its terms of use describe when ElevenLabs may use content to improve its models and offer an opt-out.
- OpenRouter (planning and writing help): OpenRouter keeps prompts only when prompt logging is switched on for the account; the policy of the model provider it routes to also applies.
3. Output responsibility
AI outputs can be inaccurate, inconsistent, or legally risky. You are responsible for reviewing what you publish, especially around likenesses, trademarks, copyright, and factual claims.
4. Provider constraints
- Providers may moderate or reject requests.
- Providers may have their own output rights, safety rules, or retention windows.
- Certain models may not support native audio, references, or extension and may change behavior over time.
- Some video and audio features are powered by Runway.