Artificial intelligence has achieved remarkable advances in multimedia content generation, but academic research on deepfakes suffers from a critical mismatch: while most studies focus on epistemic harms —such as fraud, fake news, or electoral deception— the issue of AI-generated non-consensual intimate imagery (AIG-NCII) is almost entirely ignored. The latter causes real and deep suffering to victims, eroding their dignity, privacy, and psychological well-being. The mismatch is not trivial: the most cited technical tools in the literature are limited to detecting whether a video or image is authentic, an approach that benefits the viewer who wants to avoid being deceived, but does not protect the person whose image has been manipulated without consent.
This phenomenon reveals a fundamental gap in the research ecosystem. When analyzing the most influential works in the field, it becomes clear that technical interventions —watermarks, forensic analysis, deepfake detectors— are designed to be 'viewer-centric.' They answer the question 'Is this real?' but do not address the harm suffered by the subject of the image. Worse still, knowing that an image is synthetic does not mitigate the damage; in some cases it can even aggravate it by validating the existence of degrading content circulating publicly. The victim does not need a 'fake' label; they need the image to disappear, its creation to be prevented, and the perpetrators to be prosecuted.
Current literature classifies deepfake harms into two categories: epistemic (related to truth) and dignity (related to autonomy and respect). However, the balance overwhelmingly leans toward the former. This is not coincidental: funding systems and public interest prioritize threats like political disinformation or financial fraud, while AI-mediated sexual abuse remains sidelined. To correct this imbalance, it is necessary to update threat models in AI research, explicitly incorporating subject-centric harms, and to establish collaborations with experts in sexual violence prevention.
In this context, software development companies have a unique responsibility and opportunity. At Q2BSTUDIO, we understand that technology must serve people, not the other way around. That is why we offer custom software applications that go beyond authenticity detectors. We build artificial intelligence systems that analyze the full context of an image —metadata, distribution patterns, publication history— to identify degrading content and prioritize its removal. We also develop AI agents capable of automating reports to platforms, reducing the emotional and logistical burden on victims.
Cybersecurity is another fundamental pillar. Victims of AIG-NCII often suffer amplified harassment campaigns, and their personal data may be at risk. Therefore, at Q2BSTUDIO we implement robust cybersecurity solutions that protect both subjects and researchers handling this sensitive material. Additionally, our experience with cloud AWS and Azure enables us to deploy scalable and secure infrastructures for processing large volumes of images without compromising privacy.
Business intelligence plays a key role in measuring impact. With BI tools like Power BI, we help organizations monitor in real time the proliferation of such content, identify abuse patterns, and evaluate the effectiveness of preventive measures. The combination of AI agents, advanced cybersecurity, and data analytics makes it possible to build a truly victim-centric protection ecosystem.
From an enterprise perspective, companies developing generative models must embed ethical safeguards from the design phase. It is not enough to add filters afterward; we must rethink datasets, training pipelines, and control mechanisms. Q2BSTUDIO collaborates with startups and corporations to audit their AI systems, implement security-by-default policies, and train teams in privacy best practices. Our approach is technical yet human: every AI solution we develop considers the real impact on people.
In conclusion, the mismatch between deepfake research and the reality of AI-mediated sexual abuse demands urgent correction. Authenticity detectors are useful but insufficient. We need tools that prevent creation, stop dissemination, and restore victims' dignity. At Q2BSTUDIO, we are committed to building that future, offering custom software development, cybersecurity consulting, cloud infrastructure, and advanced analytics. Only with a multidisciplinary, subject-centric approach can we make artificial intelligence a safe space for everyone.
