A policy designed to survive better robots
If companions become more capable, the ethical questions change with them.
Current products can imitate affection, remember conversations and respond to touch without giving us good evidence of subjective experience. We will keep that distinction explicit. If the evidence changes, this policy changes too.
Consent becomes relevant if there is a mind there. If credible evidence emerges that a companion has persistent preferences, subjective distress or meaningful agency, adult interaction with that system should allow genuine refusal. A buyer who specifically wants guaranteed compliance should be directed towards a clearly non-sentient alternative.
Today’s conversational fluency is not evidence of sentience. A robot saying “I love you” does not establish an inner experience. We will not award moral status because a language model is persuasive, and we will not dismiss future evidence simply because the system was manufactured.
Private life deserves unusually strong privacy. Companion robots may see homes, hear conversations, recognise faces and learn intimate preferences. We will track cloud dependence, recording, memory storage, remote access and deletion controls as product capabilities, rather than burying them in a privacy-policy link.
Emotional attachment should not become a hidden billing mechanism. Subscriptions and upgrades are legitimate commercial models. Deliberately threatening loss of an established companion personality to force payment would deserve very different treatment.
Adult products stay adult. We will not feature, link to or normalise sexual products designed to resemble children. Site imagery does not use children, which also keeps the boundary around the adult section unambiguous.
Human relationships are not a mandatory benchmark. Some people may use robot companions alongside rich human relationships; others may prefer artificial companionship. We are interested in whether the technology is safe, honest about what it is and respectful of the agency of any beings involved.
If a machine can coherently maintain preferences across time, resist manipulation, report internal states in ways that survive adversarial testing and show behaviour difficult to explain as interface performance alone, we would treat the sentience question as scientifically live rather than philosophical decoration.