In December 2025, a Futurism report documented an extreme but cautionary case: a woman working at a generative image startup became psychotic after repeatedly using company tools to generate variations of her own photographs.
She recalled that during that period she was online almost around the clock, repeatedly tweaking prompts and parameters to generate hundreds or thousands of facial images of herself—from highly stylized futuristic portraits to exaggerated, distorted variants. What began as product testing and art experiments gradually turned into a compulsive behavior. She wrote that as the number of generations increased, her dreams became fragmented, and she continued to "see" those unreal faces after waking, and began to suspect the motives of people around her.
When she told colleagues about her condition, she did not receive timely medical help but product‑focused advice: change the training dataset, adjust model parameters, try different regularization methods. Colleagues discussed image quality and style consistency rather than her mental‑health symptoms. After a clearly abnormal episode, her family took her to see a doctor, who diagnosed a bipolar disorder manic episode accompanied by hallucinations and paranoid psychotic symptoms.
The report quoted her and internal company sources, detailing behaviors such as repeatedly deleting and reposting images during generation, reduced social media interactions, and a complete collapse of work‑life boundaries. She described how the AI‑generated "half‑real" faces began to disrupt her sense of reality—the boundary between reality and algorithmic outputs blurred, becoming an important trigger for emotional and cognitive disturbance.
Futurism did not simply attribute the case to the technology itself but pointed to several risks: first, generative AI products lack mechanisms to identify and warn about "self‑image addiction" risks; second, startups often neglect employee mental‑health support amid rapid iteration; third, current regulation and ethical guidelines do not yet cover the mental‑health risks posed by internal testing and repeatedly generating images of the same subject.
Several mental‑health experts interviewed warned designers and operators of generative AI to address two areas: incorporate usage time limits, warning mechanisms, and offline‑switch options into product design; and establish mental‑health support channels at the company level, with management procedures when employees handle sensitive materials (such as their own face or private images). Experts also noted that some individuals already at risk for mood disorders may be triggered by algorithmic "mirror feedback."
The incident sparked discussion in technology and mental‑health circles over how technology availability and frequency of exposure may become new public‑health variables, and how companies should balance innovation with employee protection. The report concluded that as generative AI increasingly enters personal life, social media, and creative workflows, related ethical standards, product safety mechanisms, and workplace mental‑health protections need rapid development.
In this December 2025 case, the interaction between individuals and technology is amplified into a real warning: when AI can continuously and precisely reproduce human images, the technology brings not only creativity and convenience but also potential new psychological risks.
▶️ Enjoy this content? Creating is hard!
Subscribe to our YouTube channel:
AI Rosetta
Your likes, subscriptions, and shares are our greatest support! 🙏
For more AI news, visit: https://www.airosetta.com