Loading RunAsh
Our commitments to creators, viewers, and society — with specific, auditable requirements, not vague promises.
Core commitments
Every AI decision in a live session can be reviewed, overridden, or disabled by the streamer in real time. Automation is a tool, not a replacement for human judgment.
We publish model cards, training data summaries, and bias audits for every production model. Viewers are always notified when AI is active in a stream.
We collect only what is necessary, retain nothing beyond session scope by default, and never sell viewer data to third parties.
Our models are tested across socioeconomic contexts and regional dialects. We actively fund research into underrepresented commerce verticals.
A dedicated AI Ethics board reviews policies quarterly. Incident reports are published publicly within 30 days of resolution.
We prohibit AI features that exploit psychological vulnerabilities — scarcity loops, fake urgency signals, or manipulative persuasion patterns.
Explicit prohibitions
Generating false scarcity or countdown timers with no real inventory constraint
Synthetic endorsements using real person likenesses without explicit consent
Targeting minors under 16 with commerce-optimised CTAs
Collecting biometric data (facial emotion, gaze tracking) without explicit opt-in
Using AI to impersonate a human host without disclosure to viewers
Manipulative dark patterns designed to prevent cart abandonment via emotional pressure
Training on user conversation data without explicit written permission
Deploying AI in health, medical-device, or financial-advice contexts without human review
Regulatory compliance
GDPR
European Union
CompliantCCPA
California, USA
CompliantDPDP 2023
India
CompliantPDPA
Southeast Asia
In ProgressQuestions answered
Report policy violations, submit bias reports, or contact our Data Protection Officer.