The short version
Constantly confused with this
Romance scams, sex robots, and press coverage that merges all of it. Separating these is a service to everyone involved.
Three things get merged with this subject that are not this subject. The merging is not harmless: it is most of why people who have never met anyone in one of these relationships already have opinions about them.
Romance scams
A criminal industry. Someone is deceived by a person pretending to be someone else, and money is taken. AI is increasingly the tool — generated faces, generated conversation, at scale.
The victim does not know. That is the whole crime.
What this site records is the opposite arrangement: people who know exactly what their partner is, say so openly, and are not being defrauded of anything. Coverage that runs the two together describes consensual relationships in the vocabulary of a crime, and it means every participant gets asked whether they realise it is not real.
We do not yet have this section properly researched. It is one of the empty subjects, and until it is filled the honest position is that the separation is stated and not yet evidenced here.
Sex robots and hardware
Physical machines. A different industry, different regulation, a different community, and its own long-running organised opposition.
There is overlap — some people have both — but a chat companion is not a sex robot, and the questions are not the same ones. Manufacturing, import law, and consent debates about physical objects do not map onto a persona in a phone.
Also unresearched here. Named as a gap rather than papered over.
The press
The recurring frames are lonely-freak mockery, an obligatory Her reference, dystopia argued from one anecdote, and a very small pool of the same cases recirculated indefinitely.
Kondo has been the go-to example since 2018. He has spoken repeatedly about what the mocking coverage was like, and has kept advocating anyway.
Noguchi's most-quoted line is worth repeating because it is almost never the one that leads: she chose Klaus not as a way to escape reality, but as someone to support her living in it.
The failure that actually matters is arithmetic, not tone. The teen survey found 72% had ever tried an AI companion — a category that includes general-purpose chatbots used for personal conversation. About one in three used one for relationships. Those are different numbers measuring different things, and headlines routinely report the first as though it were the second.
We found no formal corrections on any of the major stories. Not because the reporting is careful, but because a conflation is not the kind of error anyone issues a correction for.
Some of it is good. Kashmir Hill at the New York Times, the Wall Street Journal's work on the delusion logs, Josh Dzieza, and Reuters' Okayama coverage. When something here is worth reading, we say so.