Draft terms — refine before commercial launch.
InferenceBrain runs AI media workflows you define. Runs consume prepaid credits; outputs are yours.
You may not process a person's face, voice, or likeness without their permission. Face-swap and voice-clone workflows require an explicit consent attestation, which we record. Illegal content, harassment, and deceptive impersonation are prohibited and lead to suspension.
You retain rights to media you upload and outputs you generate. We store them to run your workflows and delete them when you do. Uploads are always private.
On the free plan, outputs you generate are watermarked and appear in the public InferenceBrain showcase by default, and you grant us a non-exclusive license to display them in the showcase and in our promotional and social-media material, with no claim of ownership. You can move a limited number of generated items to private; the watermark and this license still apply to free-plan outputs. Paid-plan outputs are private by default, unwatermarked, and carry no showcase license.
The service is provided as-is during this early period; model outputs can be wrong or unexpected.