Joi AI is trying to frame erotic AI companions as more than entertainment. In a 28-day in-house experiment, the company asked paid participants to test whether AI-guided masturbation could function as a structured wellness ritual.
The project produced striking self-reported numbers on stress, focus, and cravings. It also raised a harder question: when an AI companion company studies its own product, how much should anyone read into the results?
What Joi AI asked participants to do
The experiment involved 10 people described as “masturbation consultants.” They were instructed to masturbate every day except Sundays throughout July. Twice a week, they used Joi AI’s digital companions; on the other days, they used their own preferred methods.
Before and after each session, participants logged personal signals including mood, energy levels, cravings for various vices, and whether they were procrastinating. The company said it wanted to examine sleep, stress, craving intensity, focus, ease of arousal, and how AI-guided self-pleasure compared with other methods.
One participant, Keshav, a single, 38-year-old Indonesian trader, said the AI interactions could feel socially convincing. “It's almost like you are chatting with a real person,” he said. He did not want to use his surname for professional reasons.
Joi AI’s platform lets paying users chat and play with premade characters or create their own companions. The source article describes the product as including X-rated kinks such as blackmail fantasy, gangbangs, and emotional dependency, along with capabilities and functions like squirting and roleplay.
The business behind the experiment
The participants were not casual users pulled quietly into a background study. Keshav and the nine other consultants were set to receive $2,000 each for the qualitative project.
The company is using the work to promote a daily guided masturbation tool described as “jerk off with purpose,” combining “sound [and] ritual.” Joi AI says this fits into its stated aim to “solve male loneliness.”
The experiment also functioned as a public-facing campaign. More than 150,000 people applied for the 10 places. Five were chosen by the company, while five came through public competitions, including one judged by porn stars and another seeking the “most unhinged” fetish. The winner of that contest had a balloon fetish.
The product has multiple layers of monetization. The website costs a minimum of $3.99 a month to access. Users can also buy 10,000 “Joi neuron” coins for $99.99, and private videos from companions cost 1,000 neurons. Some digital companions requested gifts such as handbags and lipstick, which cost up to $5 in the platform’s internal currency.
What the early findings claimed
Before launching the study, Joi AI surveyed 2,500 adults. The company found that 75 percent masturbate to reduce stress and anxiety, while 43 percent self-pleasure before “a big event,” such as a date, exam, or difficult conversation.
Preliminary analysis of 134 masturbation session reports, seen by WIRED, suggested several self-reported changes:
- Participants' stress levels reduced by 25 percent.
- Focus increased by 17 percent.
- Self-reported cravings for nicotine, alcohol, junk food, and doomscrolling reduced by 44 percent after masturbation.
- AI-guided self-pleasure produced comparable satisfaction scores to established methods.
The company’s report said, “The study found no significant differences between the methods; so there is no one best way to masturbate.” That is a narrower and more cautious conclusion than some of the broader wellness framing around the project.
Joi AI also said its pre-research survey found that 37 percent of respondents are open to AI-guided masturbation or already practicing it. For a company selling erotic AI companions, that suggests a potential market. It does not, by itself, prove a health effect.
Why the evidence is contested
Julie Levin, head of brand and communications at Joi AI, connected the company’s interest to ejaculation, dopamine, cravings, and self-discipline. “You get this boost of dopamine when you ejaculate,” she said. “We wanted our consultants to track whether their cravings [for vices] are worse or better and how it affects their self-discipline.”
The company’s report went further, saying, “Masturbation may be the most accessible tool people have for managing addictions.” The source article notes that this claim is not based in science.
Nicole Prause, a neuroscientist studying human sexuality and addiction, was sharply critical of that leap. She called it “a gross overstatement from the evidence they claim to have.” Prause said she is not aware of scientific evidence that masturbation reduces vulnerability to substance abuse, and she questioned the research method.
Her concern centered on self-reporting. “Since masturbation can never be placebo—people always know if they masturbated—it is important to use markers that are less subject to bias than self-report,” she said. “They use none. Someone saying ‘I feel focused’ is just that. There are dozens of cognitive tests that could be used to support a self-report. I do not see any.”
Prause did not dismiss masturbation as irrelevant to mood. She said masturbation to orgasm is a potent method of mood management, and noted that the most “dramatic shifts” in her own study cohort come from people reporting decreased anxiety and increased relaxation after climaxing.
The bigger question for AI companions
The experiment also highlights a broader tension around intimate AI. Chloé Locatelli, a sex-tech researcher at King’s College London, was invited to participate because she is undertaking a collaborative project with Joi AI on how different demographics build intimate and erotic relationships with the platform’s characters.
Locatelli said masturbation is something she does to release stress and that she sees physical and well-being benefits from self-pleasure. But she was less taken with the AI companions than Keshav. She said the interaction helped with arousal “a little bit,” but did not fully satisfy what she was looking for.
Her caution was about dependency. As AI companions become more interactive and personalized, she worried about a user relying on them too heavily. “If it becomes a bit of a crutch [and] your only form of sexual exploration, then that's not something that I would necessarily encourage,” she said.
That is where Joi AI’s project becomes more than a provocative marketing experiment. It sits at the intersection of sex tech, AI companions, loneliness, wellness claims, and weakly tested self-improvement narratives. The reported data may be interesting, but the strongest takeaway is restraint: self-reported effects can point to questions worth asking, but they cannot carry claims the study was not designed to prove.