Q: Is my document sent to an AI model?By default, your document is not sent to AI providers. AI features are opt-in: you choose whether to enable AI and which features to use.When you use an AI feature, the content needed for that feature is sent to the relevant provider. Depending on the feature, this may include a question, notes, character descriptions, images, or script text. For example, reading your entire script with Table Read requires sending that text to a speech provider.Enabling AI does not itself purchase credits or start generation. Review each feature’s disclosure before enabling it.Q: If I use the AI features, what gets sent?
- Research: Your question, conversation history, and any included notes, sources, or script context are sent to the relevant text provider—OpenAI, Anthropic, or Perplexity. This content is not used for model training.
- Image generation: Your prompt, relevant card or note content, and any reference images are sent to Runway. Card details such as the title and character information may also form part of the prompt. Your script is not automatically included. Image-generation inputs may be used for model training.
- Table Read audio: The text you choose to have read is sent to Murf or ElevenLabs to generate audio. This may include your entire script if you choose to read it all. Neither provider uses this text to train its models under our account arrangements.
- Automatic voice assignment: Character names, descriptions, and gender information are sent to OpenAI or Anthropic to help choose suitable voices. This content is not used for model training.
- Script breakdowns and scene synopses: Relevant sections of your script are sent to OpenAI or Anthropic for production tagging and scene summaries. This content is not used for model training.
- Video generation, where available: The image you choose and its accompanying prompt are sent to Replicate.
Q: Which AI models are used?We use different models for different tasks. Current models include:
Feature | Models |
Research and text tools | Anthropic Claude Sonnet 4.5, OpenAI GPT-5.2 and GPT-5 mini, and Perplexity Sonar Pro |
Automatic voice assignment | OpenAI GPT-6 Luna, with Anthropic Claude Haiku 5.5 as a fallback |
Script breakdowns | Anthropic Claude Opus 4.6 |
Scene synopses | Anthropic Claude Haiku 4.5 |
Image generation | Runway Gen-4 Image |
Table Read audio | Murf GEN2 and ElevenLabs Multilingual v2 |
Video generation, where available | MiniMax Hailuo-02 through Replicate |
The model used depends on the feature, and model versions may change over time.Q: Where can I read the providers’ privacy policies?Our AI features use the following providers:
- OpenAI Privacy Policy
- Anthropic Privacy Policy
- Perplexity Privacy Notice
- Runway Privacy Policy
- Murf Privacy Policy
- ElevenLabs Privacy Policy
- Replicate Privacy Policy
API services can have different data-use terms from consumer products. See also OpenAI’s API data controls, Anthropic’s commercial model-training policy, and Perplexity’s API terms.Q: Can the information provided by AI be trusted?AI can make mistakes or produce inappropriate content. Check all output before using or sharing it.Treat AI responses as a starting point and independently verify factual claims and sources. Review generated summaries, production tags, images, audio, and voice assignments to make sure they are accurate and suitable for your work.