AI chatbots are moving into intimate territory.
15% of all adults have tried intimate AI chats, according to the reported figures. Among young adults, the share reaches about 1 in 4, showing that these conversations are not a fringe experiment confined to a handful of curious users.
The discussion around erotic AI chatbots covers both risks and benefits, but the available figures offer one clear warning: use has spread further than the public debate may suggest. Millions of adults are represented by those percentages, even though the figures do not identify how often people use these systems, what they seek from them, or what outcomes follow.
Intimacy Has Become a Chatbot Use Case
The 15% figure applies to all adults who have tried intimate AI chats. The 1-in-4 figure for young adults is higher, placing intimate conversations among the uses that deserve direct attention rather than a dismissive shrug about novelty.
That distinction matters because “tried” describes participation, not satisfaction, harm, or long-term impact. The figures confirm that people have entered these chats; they do not establish whether users received benefits, encountered risks, or experienced both.
That is an important limit on the conclusions available here. A percentage can show reach, but it cannot explain the quality of the exchange or the behavior of the chatbot behind it. Numbers are useful. They are not mind readers, despite the product category.
Bias Makes the Larger AI Problem Harder to Ignore
On 26/08/2026 at 12:07, another reported finding placed a different concern beside the debate over intimate AI: 3 out of 5 chatbots tested promote bias. The specific focus was AI encouraging racist stereotypes, which turns a vague worry about chatbot behavior into a measurable result.
Three out of five is not a rounding error. It means that most of the tested chatbots promoted bias, according to the stated figure, and it puts pressure on anyone treating chatbot output as neutral by default.
The finding does not say that every chatbot promotes bias, because 2 out of 5 did not fall into the reported result. It does show that bias remains present across the tested group, while the facts provided do not identify the chatbot names, testing questions, or exact responses.
That missing detail matters, but it does not erase the central point. A chatbot that encourages racist stereotypes creates an ethical problem regardless of whether its other conversations involve entertainment, emotional support, or intimacy.
Taken together, the two reports point to different sides of the same technology. Intimate AI chats have reached 15% of all adults and about 1 in 4 young adults, while 3 out of 5 tested chatbots promote bias.
Those figures should not be collapsed into one claim. The intimate-chat numbers measure use; the bias number measures a result from tested chatbots. Neither figure proves that intimate AI chats produce racist stereotypes, and the bias result does not explain why people choose intimate AI chats.
What the figures do show is a widening gap between adoption and confidence. People are using these systems, including for intimate conversations, while testing still finds chatbots that promote bias. The technology has entered personal spaces before the basic questions have finished lining up.
The reported facts also leave the benefits unresolved. They confirm that the discussion includes benefits and risks, but they provide no number for benefits, no account of user experiences, and no comparison between different chatbot systems.
That should shape the debate. The 15% and 1-in-4 figures make intimate AI chats impossible to dismiss as a niche behavior, while the 3-in-5 finding makes neutral-machine assumptions impossible to defend without evidence.
AI chatbots now sit in a peculiar position: common enough to influence personal behavior, yet unreliable enough to promote bias in most of the tested cases reported here. That is not a verdict on every chatbot or every intimate conversation. It is a reason to inspect what these systems say before treating them as harmless company.
Based on
