Where we use AI
Voice generation and cloning, speech to text, video and image generation, clip finding and captions, translation, writing help, our assistant King, and chat's @Edge helper. Models run on GPU machines we operate and, for some tasks, through providers listed on Subprocessors.
Labelling
- Audio and video we generate carry hidden provenance: an inaudible watermark and metadata saying "AI-generated" and the job. Do not remove them. Removing them breaks Acceptable Use.
- When you publish AI-generated media, follow each platform's rules on labelling AI content.
- King always says it is an AI assistant. Chat messages from @Edge are marked as assistant messages.
Limits
- AI output can be wrong, biased, repetitive or unsuitable. Voices for some African languages have a limited range and may sound neutral or make mistakes.
- Do not use output as legal, medical or financial advice.
- Generated voices can sound like real people. Cloning requires the owner's recorded consent (Voice consent).
Your content and training
We do not use your content to train our models, and we do not allow our providers to train on it, unless you separately opt in.
Automated decisions
We use an automated risk score for fraud prevention (one account per person, trials, referrals). It never blocks someone on one signal alone. You can ask a person to review any decision at [email protected].
Moderation
Prompts, uploads, outputs and chat messages are checked automatically, with human review of flagged items.
Commercial rights
Depend on your plan and model. See Content licence.