New Study Shows AI Deepfake Abuse Increasingly Targets Everyday People Beyond Celebrities
A survey of 7,230 adults across three nations reveals that non-consensual AI deepfakes now widely affect ordinary citizens, prompting calls for stricter developer guardrails and new legal protections.
A newly published international study highlights how AI-generated image-based sexual abuse has expanded beyond public figures to impact everyday individuals across the globe. Conducted in July 2024 with 7,230 adult participants from Australia, the United Kingdom, and the United States, the research reveals that 7 percent of respondents have experienced non-consensual sexual deepfakes. The data underscores how accessible, consumer-facing generative tools have lowered the technical barrier to creating harmful synthetic media, shifting the deepfake phenomenon from high-profile targets to the broader public.
The findings show that demographic groups including people under 35, LGBTQIA+ individuals, Black, Indigenous, and other people of color, individuals with disabilities, and men reported higher rates of victimization. However, women reported a wider array of negative emotional and social consequences, such as severe distress, embarrassment, and fear that synthesized images could harm their professional prospects. Furthermore, data from the Australian eSafety Commissioner shows a steep rise among younger demographics: reports from children regarding digitally altered intimate media more than doubled in the 18 months leading up to June 2025 compared to the preceding seven years combined, with 80 percent of those cases affecting girls.
Harms and motives behind synthetic abuse
Beyond targeted harassment or revenge, the study revealed that peer culture and casual experimentation drive much of the creation and spread of non-consensual intimate media. Approximately 3.2 percent of survey respondents admitted to creating, sharing, or threatening to share non-consensual AI intimate images. Perpetrators cited motivations ranging from wanting to experiment with technology or show off to peers, to viewing the behavior as harmless humor or social bonding.
Despite these trivialized motivations, victim-survivors face significant psychological harm and administrative burdens. Many experience a chaotic cycle of searching for and attempting to take down circulating images, only to find them republished elsewhere. Additionally, victims frequently encounter skepticism from law enforcement and personal acquaintances who dismiss the abuse because the synthetic images do not depict real photographic capture. One UK participant noted that authorities and peers demonstrated skepticism because the material was not deemed a "real image," leading to feelings of isolation and helplessness.
Regulatory actions and platform responsibilities
In response to the surge in non-consensual synthetic media, legislative frameworks are beginning to target both perpetrators and the technical ecosystem enabling them. Australia recently introduced a proposed digital duty of care in parliament, which would require digital services—including AI image generation tools—to implement proactive measures to protect users. The legislation also aims to empower Australia's eSafety Commissioner to directly target "nudify" applications and websites that digitally strip clothing from uploaded photos.
Other jurisdictions are taking even stricter measures. The United Kingdom has moved toward criminalizing the actual development and supply of automated nudify applications. Researchers advocate for comprehensive criminal offenses covering both the creation and distribution of unauthorized synthetic intimate media across all jurisdictions, combined with consent-focused educational programs to counteract the normalization of digital abuse.
What it means for developers
For software engineers, product teams, and model providers, these findings emphasize the urgent requirement to integrate safety mechanisms directly into model architecture and API endpoints. Rather than relying solely on post-hoc moderation or placing the reporting burden on victims, platforms are expected to implement strict input-prompt filtering, robust output guardrails, and rapid removal workflows. Build-time precautions must prevent AI models from generating non-consensual sexual content or processing reference images without clear consent verification.
As regulatory pressures such as digital duty of care statutes increase, developers building on generative AI infrastructure must evaluate model safety and compliance features carefully. For developers building consumer applications or testing multi-model capabilities, accessing top AI models cheaply through one API at https://apixoai.online offers a streamlined way to integrate and test models with varying guardrail implementations and safety features.
Ultimately, preventing synthetic image abuse requires proactive system design, strong API guardrails, and strict compliance with emerging international legal standards to keep end users safe from unauthorized AI manipulations.
Source: ‘Nobody cared’: new research shows deepfakes are hurting regular people, not just celebrities — The Conversation AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
