Trust
Responsible AI at 3Stone
3Stone can help people reason, research and create. It can also be wrong. Important work should be reviewed in proportion to its impact.
01
Limits
Models may generate inaccurate, incomplete or biased material. Current information can change. Generated media can look real without depicting a real event.
02
Safeguards
3Stone uses provider and application safety controls, durable execution boundaries and visible failure states. Consequential actions require the relevant authorization rather than being inferred from a conversation.
03
Your role
Review high-impact outputs, respect rights and consent, do not use generated content to deceive, and report harmful or materially incorrect behavior through the product or contact path.