AI romance cheating: 50.5% say yes, and users judge it hardest

Half of American adults, 50.5% of them, say a partner's romantic or sexual relationship with an AI can count as cheating. The headline number is not the interesting one. Further down the same table sits the finding that people who currently use AI companions call it cheating at 80.6%, while people who have never used one say so at 44.0%. The strictest judges here are the ones doing it.

Who asked, and who paid

The survey ran on 26 and 27 August through SurveyMonkey Audience. It started with 2,150 US adults and kept 1,709 after screening, which is a normal size for this kind of work and a normal opt-in online panel rather than a probability sample.

It was commissioned by AI Girlfriend Coach, a company that sells help using AI companions. That does not make the numbers wrong, and it does mean a reader should want the result confirmed by someone with nothing riding on it.

The number holds up elsewhere

Independent work lands in the same place. A Canadian team surveyed 1,815 adults and found roughly half treating a partner's romantic AI use as infidelity, with about three quarters reporting a negative emotional reaction to the idea. The splits inside that sample are sharper than the average: cisgender women were twice as likely as cisgender men to call it cheating, Gen Z respondents reacted more harshly than older ones, and people in non-monogamous relationships were half as likely to see infidelity at all.

“The key question is not simply whether romantic companion use counts as cheating, but how it develops, whether it is disclosed, what needs it serves.”

— David Lafortune, Professor of Sexology, UQAM, The Conversation, 6 July 2026

Quote source: The Conversation, 6 July 2026

The detail most readers will skip

The detail most readers will scroll past is who actually uses these products. Among people engaged or married, 30.7% have used an AI companion. Among single people, 16.3% have. The behaviour is roughly twice as common inside a relationship as outside one, and it is the only figure in the survey describing what people do rather than what they think about it.

What people say they are there for, in the same survey:

• companionship or loneliness: 56.6%
• affection, romance or sex: 10.3%
• some limit on AI intimacy: 71.7%
• no concern, because an AI is not a real person: 28.3%

Read those two blocks together and the disagreement stops being about technology. Most users describe a loneliness problem. Most partners hear an intimacy problem. Both descriptions can be accurate at once, which is why the argument is hard to end and why 21.2% of respondents refused the word cheating while staying uncomfortable anyway.

Regulators have already started drawing the line for their own reasons. China switched off companion features in Doubao and Qwen in July under a new law. The commercial pressure runs the other way, and the same emotional pull that makes a companion sticky is the one that romance scammers have monetised for years.

The number to watch is not the 50.5%. It is whether the 30.7% keeps climbing, because a norm argued about by half the country tends to get settled by whoever is already living inside it.

None of this should be read as personalized investment advice.