Many individuals developing a sexual interest in minors are drawn to images of AI children on TikTok and Instagram
However, when such content is lawful and portrays fabricated individuals, it enters a complex and concerning ambiguity
The young ladies depicted in the images shared on TikTok and Instagram appear to be no older than five or six years. Beyond the age of thirteen, on the elder end.
They are depicted donning leather and lace, crop tops, and bikinis. They are sensuously attired in the personas of ballerinas, nurses, French ladies, and superheroes. Others prefer pigtails and oversized spectacles to rabbit ears or devil horns. They are blonde, brunette, and black, as well as Asian and Caucasian.
They were all created using artificial intelligence, and they have attracted the interest of an unsettling demographic on some of the most popular social media platforms worldwide: older males.
One TikTok user commented on a recent post of young blonde girls dressed as maids with flowers and ribbons around their necks, “AI creates magnificent works of art; I would love to procure a delicate virgin like that to make it mine.”
Does it make me look bad if I say she is hot in every way if this is artificial intelligence-generated? A separate TikToker commented on a slideshow featuring completely attired young girls in Spider-Man costumes. “She does not exist, but my imagination is quite vivid.”
“In any case, it shouldn’t be incorrect,” responded another whose username included “p3do,” which could be coded language for “pedo” or algorithmic language.
Another stated, “You cannot violate her because it is an AI image.”
Similar comments inundated Instagram with images of AI children. Someone wrote, “I would love to take her innocence even if it’s a false image,” in response to a post featuring a petite, pale child attired as a bride.
The same user commented on another image of a young girl wearing short shorts, comparing “her adorable pair of small size [breasts]” (represented by two apple emojis) to “her perfect innocent slice of cherry pie down below.” “Gary” commented on a photograph of a preteen wearing garter straps, remarking, “Nice little buds. As she matures, these will develop into sizable mangoes.”
“It appears to be tasty.” “Do home deliveries occur?” “At the optimal age to be exploited.” “I desire that adorable baby.” “Can you conduct a test in which she leaps from my bed to my phone?” inquired another user on TikTok and Instagram. Forbes discovered hundreds of similar posts and remarks on the platforms beginning in 2024 regarding images of AI-generated children. Many were attached to popular songs to attract more attention, such as “Texas Hold ‘Em” by Beyonce, “Shake It Off” by Taylor Swift, and “Fast Car” by Tracy Chapman.
TikTok and Instagram have emerged as the most popular platforms among teenagers and younger users, despite the fact that child predators have explored the majority of well-known social media platforms and taken advantage of the privacy and screen space these platforms offer.
Although the difficulty platforms have in regulating child sexual abuse material (CSAM) predates the current AI growth, the use of AI text-to-image generators facilitates predators’ access to or creation of such content.
The tools have contributed to increased CSAM generated by artificial intelligence, which is unlawful even if forged. However, the images that have been revealed through the coverage of this story are ambiguous. Although not overtly explicit, they are sexualized. Minors are depicted, but not actual ones.
Despite their apparent legality, the remarks expressed regarding them imply malevolent intentions. Child protection and forensics experts have characterized them as entrances to considerably more sinister and potentially unlawful operations.
This causes tech and law enforcement to ponder how to combat the scourge of suggestive, fabricated images of children who do not exist; if such images are not explicit and lawful, the question remains whether they should be addressed at all.
By federal legislation, technology companies are obligated to notify the National Center for Missing and Exploited Children, a nonprofit organization that provides law enforcement with information regarding CSAM and child sexual exploitation suspected to occur on their platforms.
However, they have no legal obligation to flag or remove the images detailed in this article. Notwithstanding their legality, NCMEC told Forbes that social media companies should remove them.
“These are all being trained using images of real children, whether or not they are exact replicas, so NCMEC doesn’t see that side of the argument that this is acceptable in some other world,” said Fallon McNulty, director of the organization’s reporting hub for tech companies, CyberTipline.
Stable Diffusion 1.5, one of the most widely used text-to-image generative AI tools currently available, was discovered to have been trained on CSAM, or Child Social Media, using data taken from the web.
“Considering some of the commentary surrounding the images, it does not appear that the audience that is consuming, reviewing, and ingesting them is innocent,” McNulty continued. “I would like to see them remove that content and prevent individuals from sexualizing children on their platforms,” she said.
The AI-generated children’s images were tagged with Taylor Swift, Beyonce, and Tracy Chapman’s songs to increase their exposure.
In response to a request from Forbes, TikTok and Instagram permanently removed the accounts, videos, and comments referenced in this article; both platforms claimed the content violated platform policies.
“To protect young people and maintain an inhospitable environment for harmful behavior, TikTok has stringent policies against AI-generated content of minors,” said a spokesperson, Jamie Favazza.
The organization launched a novel synthetic media policy on Friday, which strictly forbids such material if it features an individual under 18. In addition, posts that violate the policy are removed without regard to whether they were modified using artificial intelligence.
A spokesperson for Meta, the parent company of Instagram, Sophie Vogel, stated that the platform prohibits and removes accurate and AI-generated content that exploits or sexualizes minors.
Vogel noted that they continue to remove profiles, accounts, and pages devoted to sharing images of children or remarking on their appearance, even when the content appears harmless. Meta and TikTok both notify NCMEC of AI-generated CSAM discovered on their platforms.
A ‘Gateway’ To Illicit Material
A well-known creator of AI-generated children amassed 80,000 followers on Instagram and TikTok; that number increased by the thousands as Forbes reported this story.
The Mandarin-language bio of the account read, “Woman With Chopsticks.” Forbes encountered difficulties ascertaining the operator of its most ubiquitous posts; however, they garnered hundreds of comments, were saved or shared thousands of times, and were viewed close to 500,000 times.
Based on their profile pictures, handles, bios, and comments, many of their followers appeared to be older males.
Heather Mahalik Barnhart, an authority on digital forensics, stated that in her previous investigations into these “child erotica” cases, the followers frequently provided investigators with glaring indicators that could have identified potential perpetrators. “Child erotica is the gateway,” she told Forbes, adding, “It’s not normal to see who is following it.”
Additionally, potential offenders are more likely to be exposed through these accounts because they feel more comfortable being exposed in public.
“People feel more secure because the children are not completely exposed; therefore, there is no CSAM,” said Mahalik, who oversees the software company Cellebrite’s efforts to assist law enforcement and NCMEC in their investigations of child exploitation using its technology. She emphasized that law enforcement must scrutinize the individuals who follow these accounts more closely and search for behavioral patterns.
A tutorial video was published on the TikTok account, demonstrating how to create and refine their photographs featuring young females, including details such as their teeth.
“I fell in love with you the moment I saw your eyes, even though we are many years apart,” one TikTok user wrote. The user was following other accounts that posted even more explicit images of children. (An additional frequent commenter, whose username was “open wide,” maintained an account where she featured photographs of actual women stretching in leotards and performing splits.)
Certain commentators on the TikTok and Instagram posts authored by “Woman With Chopsticks” inquired about the AI model employed in creating the AI children, implying a potential interest in further development.
This stuff is incredible; I wish I could find an app to make it. (I wrote about one man on Instagram a week ago, alongside an image of a young blonde wearing only a bralette and minimal clothing.) The TikTok account published a three-minute instructional video in January that demonstrated to users how to create and refine their photographs featuring young females, including details such as their teeth.
Although the creator designated some of the posts “AI-generated,” as TikTok requires, it can be difficult for the unaided eye to discern that they are unreal. This complicates matters for law enforcement in their efforts to identify actual victims. (A spokesperson for Meta, Vogel, states that the company is developing tools capable of discerning images created with OpenAI, Midjourney, Microsoft, Google, Adobe, and Shutterstock. Subsequently, the tools will be employed to designate AI-generated content shared on platforms including Facebook, Instagram, and Threads.)
Director of Information Technology at the Canadian Centre for Child Protection, Lloyd Richardson, stated that these borderline images, whether created artificially or manually, serve as a “gateway” to more severe or illegal material, frequently on other platforms, which poses “a clear safety risk to children.”
“The fundamental concern is that this subset of images provides opportunities for criminals to network,” Richardson told Forbes, adding that they frequently transfer their discussions to private direct messages. “By serving as signposts, these images can direct attention to CSAM through other channels.”
A slideshow on TikTok on January 13 depicting young females dressed in silk pajamas exemplified this type of back-and-forth in action. Someone wrote in the comments, “This works for me.” In response, another replied, “I just sent you a message.”
Richardson continued, “This is why organizations cannot moderate content in isolation by examining images alone.” “They must be considered in the greater scheme of things in terms of how they are viewed, shared, and followed.”
This is undeniably more difficult, as contextual evaluation may be necessary, necessitating the involvement of human moderation staff. (A spokesperson for TikTok, Favazza, stated that automation is utilized to identify potential predatory conduct.)
Those who have a sexual interest in children can discover even more images of this nature with the assistance of TikTok’s robust algorithm. In tandem with Forbes’ coverage of this story, TikTok started suggesting additional prompts, such as “ai-generated boys” and “AI fantasy girl.”
TikTok’s “For You” channel, which appears upon app launch and presents viral videos that the company predicts users will enjoy, guarantees an expanded audience of one billion.
One of the most significant risks associated with AI children sharing these viral moments on social media, even if the images are noncriminal, according to McNulty of NCMEC, is that individuals may lose their sensitivity to how harmful they can be.
“As a society, are we simply going to accept this content without question?” she inquired.
