AI Transparency
Everything VINC renders is made by artificial intelligence. This page says so plainly, explains how, and tells you what that means for you.
You are working with AI
The VINC studio — including the Director chat and the crew agents — is an AI system. The Director is an AI assistant, not a human — if you ask it whether it is human, it will tell you plainly that it is not. Every image, video clip, voice, and music cue the studio produces is AI-generated.
Your content is not training data
VINC does not train, fine-tune, or substantially modify generative AI models, and does not use your prompts, uploads, or rendered films as training data. We orchestrate third-party models under license; each provider is contractually restricted to processing your content to deliver your render. (This is also our disclosure posture under California's Training Data Transparency Act, Civ. Code §§3110–3111: VINC is not a "developer" of the models it orchestrates.)
The models behind your films
VINC orchestrates third-party and first-party machine-learning models to render your work. Depending on the shot, your prompts and reference media may be processed by: large language models (Anthropic's Claude family) for direction and planning; vision models (Google Gemini) for quality control; image models (Midjourney, GPT-Image, and others); video models (Seedance, Kling, LTX, and others via fal.ai); and audio models (ElevenLabs and others) for voice, music, and sound. The exact model mix evolves as better ones ship — the categories above are kept current in this document.
How AI content is marked
- Where technically feasible, rendered output carries machine-readable provenance identifying it as AI-generated (metadata / content-credential signals such as C2PA, as pipeline support matures). Removing or falsifying these signals violates the Acceptable Use Policy.
- AI-generated media we publish ourselves — like the films on our landing page — is labeled as AI-generated.
Your duties when you publish (you are the deployer)
Under the EU AI Act, when you publish content made with VINC, you are the deployer of that content and its transparency duties fall on you. In particular, content that realistically resembles real people, places, or events (a "deepfake" in the Act's language) must be visibly disclosed as artificially generated where you publish it — with a lighter, non-intrusive acknowledgement allowed for evidently artistic, creative, satirical, or fictional works. India's synthetic-media rules and several US state laws impose similar labeling duties. Our Acceptable Use Policy makes this labeling contractually mandatory, and deceptive deepfakes of real people are banned outright.
Limitations you should know
- Generative output is non-deterministic: the same prompt renders differently each time, and creative quality varies.
- Outputs can be inaccurate and can unintentionally resemble real people, places, or copyrighted works. Review before you publish.
- Automated safety filters screen prompts and outputs (see Terms §6 for moderation and appeal). They are imperfect in both directions — a wrongly refused render can be appealed via the grievance process.
Questions
Write to support@vinc.ai. For how your prompts and media are processed as personal data, see the Privacy Policy.