Ad description

A paid-for Facebook ad for AI Realities, seen on 13 May 2026, included the caption “[heart emoji] Create, customize [sic], and enjoy limitless fun with AI girls”. 
 
The post included a video of a woman speaking and on-screen text that stated, “That’s it AI has officially gone too far I genuinely don’t think anyone can even tell that I’m not real”. 
 
Another woman then appeared in the corner of the screen in a pink outfit and with pink hair, whilst another version of her appeared in the centre of the video. The woman in the corner of the screen stated, whilst similar text appeared on screen, “Create your perfect AI girlfriend for free. You choose how she will look. No limits, text her, call her, control her. She will do anything for you. She never ignores and always wants your attention. Create her for free on Candy AI”. 
 
The video included a demonstration of how the app worked and scrolled through images of women whilst showing options to “Create my AI Girl” and select whether they were “Realistic” or “Anime”, and their “Ethnicity”, “Age”, “Eye Colour”, “Hair Style”, “Personality” and “Relationship”, before cutting back to the woman in the centre of the screen who was moving around and posing.

Issue

The complainant, who believed that the ad objectified women, challenged whether it was offensive, irresponsible and promoted a harmful gender stereotype. 

Response

EverAI Ltd t/a AI Realities said the ad focused on the capabilities of their AI technology and selected traits for an “AI girlfriend” character while also presenting wide-ranging options for personality options. It did not contain sexualised imagery of any kind. The ad highlighted how realistic their AI-generated images and videos were, how highly customisable the characters could be, and the product’s messaging and voice call functions, emphasising that exchanges with AI characters were customisable and personalised. 
 
They said the term “control her” was used to mean “customise” or “individualise”, referring to the technical act of customising an AI character’s attributes via on-screen menus. There was nothing in the ad’s imagery or language to suggest that women should be controlled, that controlling women was desirable, that women should be submissive, or that the ad referred to coercive or submissive control. The ad did not show a woman catering to a partner’s desires, or suggest that attentiveness or availability were fixed, female specific, or necessary qualities in an attractive partner. The fact that other personality options were shown conveyed that a range of qualities could be selected in “your AI girlfriend”. 
 
The ad presented detailed personality options, which acknowledged women’s personhood and agency, and presented that positively. They did not believe that acknowledging partner preferences, in the context of creating a fictional character, fed into negative gender stereotypes or domestic abuse. They also said they advertised the same fully customisable product to all sexual orientations and gender expressions, and that the message of the ad was not dependent on the protagonist being a woman or on the specific personality depicted in the ad. 
 
They acknowledged that the character in the ad would be considered classically attractive, and that attentiveness and sensitivity were traits often associated with women. However, they said that was not inherently problematic, particularly because the ad also presented other non-stereotypical traits and options. 
 
They said they had removed the ad and would amend the claim “control her” to “control your experiences”, or wording to that effect.

Assessment

Upheld 

The CAP Code stated that ads must be prepared with a sense of responsibility to consumers and to society, must not cause serious or widespread offence and must not include gender stereotypes that were likely to cause harm. 
 
The ad was for the Candy AI app, which the ASA understood enabled users to create an AI companion. The ad included a demonstration of the app being used to "Create my AI Girl". Users were shown a range of options to customise the AI character, including whether they were realistic or anime-style, as well as characteristics relating to their appearance, such as ethnicity, age, hair colour and eye colour, and aspects of their character, including personality and relationship type. 
 
The ad included the claims “enjoy limitless fun with AI girls” and "Create your perfect AI girlfriend […] control her. She will do anything for you. She never ignores and always wants your attention". While the ad promoted an AI-generated character rather than a real woman, we considered that consumers were likely to understand the "AI girlfriend" as a representation of an idealised romantic partner, both in terms of how they appeared and how they behaved. 
 
We noted EverAI’s view that the claim “control her” was intended to refer to the ability to customise the AI character’s attributes. However, we considered that because the claim was immediately followed by “She will do anything for you. She never ignores you and always wants your attention”, consumers would understand it as referring to having control over the AI girlfriend’s behaviour. We further considered the ad portrayed the AI girlfriend as desirable specifically because her physical attributes could be selected, she could be controlled, was permanently available and would always comply with the user's wishes. By presenting those traits as positive attributes in a female romantic partner, the ad reinforced the harmful stereotype that women should be submissive and compliant to men's needs. 
 
We further considered that the emphasis on creating a “perfect" girlfriend and controlling her behaviour presented the female partner as something that could be tailored to meet the user's preferences and used for their gratification. 
 
In that context, we considered the ad reduced women to objects whose value lay in their appearance, availability, controllability and willingness to satisfy the user's desires. We therefore concluded that the ad objectified women and included a gender stereotype that was likely to cause harm. 
 
For those reasons, we concluded the ad was harmful, irresponsible and likely to cause serious and widespread offence. 
 
The ad breached CAP Code (Edition 12) rules 1.3 (Social responsibility), 4.1 and 4.9 (Harm and offence).

Action

The ad must not appear again in the form complained of. We told EverAI Ltd t/a AI Realities to ensure that their ads were socially responsible and did not cause serious or widespread offence, including by featuring a harmful gender stereotype by objectifying women. 

CAP Code (Edition 12)

1.3     4.1     4.9    


More on