The whispers started subtly, then exploded online. Within hours, millions had seen fabricated images and videos depicting a woman who had never consented to their creation or distribution. The abuse wasn’t limited to digital spaces. it followed her into the real world, a chilling echo of the online harassment. This is the new reality for a growing number of women and girls, victims of AI-powered deepfake technology used for malicious purposes. While the technology itself isn’t new, its weaponization against individuals, particularly women, is escalating at an alarming rate, and the systems designed to protect them are failing to keep pace.
The proliferation of deepfakes – manipulated images, audio, or videos that convincingly portray someone saying or doing something they never did – represents a dangerous evolution of online abuse. The ease with which these fakes can be created, combined with their rapid spread across social media platforms, leaves victims feeling powerless and vulnerable. The core issue isn’t simply the existence of the technology, but the lack of legal frameworks, effective enforcement mechanisms, and platform accountability to address the harm it causes. Understanding the scope of this problem, and the barriers to justice for survivors, is crucial to stemming the tide of this digital violence.
The Rising Tide of Deepfake Abuse
Deepfake technology has become increasingly accessible. Tools to create these manipulations are widely available, often free, and require minimal technical skill. According to a 2023 report, a staggering 98 percent of all deepfake videos online are pornographic, and 99 percent of those depict women. The prevalence of deepfake videos has increased by an estimated 550 percent between 2019 and 2023, demonstrating the rapid acceleration of this threat. Once posted, AI-generated content can be replicated endlessly, saved to private devices, and shared across platforms, making complete removal virtually impossible.
The impact extends far beyond the initial shock and humiliation. Harm doesn’t remain confined to the digital realm. A UN Women survey found that 41 percent of women in public life who experienced digital violence also reported facing offline attacks or harassment linked to it. This underscores the real-world consequences of online abuse, which can range from emotional distress and reputational damage to physical threats and violence.
Why Reporting Fails Survivors
Underreporting is a significant obstacle to addressing deepfake abuse. Survivors often face immense barriers to coming forward, fearing further victimization, disbelief, or blame. When they do report, the process can be retraumatizing. Survivors are frequently compelled to repeatedly view and describe the abusive content to police, lawyers, and platform moderators, often facing skeptical questions like, “Are you sure it’s not real?” or “Did you share intimate images before?”
Even if a case reaches court, the focus often shifts from the perpetrator’s actions to the survivor’s character and past behavior. Clothing, relationships, and personal history can be scrutinized, effectively blaming the victim. This systemic failure to prioritize the survivor’s experience discourages others from seeking justice and perpetuates a culture of impunity for perpetrators.
The Legal and Enforcement Gap
Despite the growing scale of the problem, prosecutions for deepfake abuse remain rare. This is due to a complex interplay of legal, technological, and logistical challenges. A key issue is that the law has not kept pace with the technology. Less than half of countries have laws addressing online abuse, and even fewer have specific legislation covering AI-generated deepfake content. Many existing “revenge porn” or image-based abuse laws were written before deepfakes existed, leaving significant loopholes.
Enforcement is further hampered by the need for specialized digital forensics expertise, cross-border coordination, and cooperation from tech platforms. Investigators struggle to collect and preserve digital evidence as content spreads rapidly and perpetrators often operate anonymously or across multiple jurisdictions. Digital forensics backlogs exacerbate the problem, causing cases to stall before they even begin. Platforms, often citing “intermediary” status, are frequently slow or unwilling to share data with law enforcement, particularly in cross-border investigations.
A Path Forward: Urgent Action Needed
Addressing deepfake abuse requires a coordinated and multifaceted approach involving governments, institutions, and tech platforms. Several jurisdictions are beginning to take action, but a more comprehensive and unified response is urgently needed.
Here are five critical steps:
1. Strengthen Legal Frameworks
Governments must enact legislation specifically addressing AI-generated abuse, focusing on consent, establishing strict liability for perpetrators, mandating fast-track removal obligations for platforms, and creating clear protocols for cross-border enforcement.
2. Invest in Justice System Capacity
Law enforcement agencies need training, resources, and dedicated capacity to collect and preserve digital evidence. Addressing digital forensics backlogs and establishing robust international cooperation frameworks are essential.
3. Hold Platforms Accountable
Tech companies must be legally required to proactively monitor for and remove abusive content within defined timelines, cooperate fully with law enforcement, and face meaningful financial consequences for failing to act. The current system of self-regulation is demonstrably inadequate.
4. Provide Comprehensive Support for Survivors
Access to trained, trauma-informed law enforcement and legal professionals, as well as free legal aid, is crucial for supporting survivors through the reporting and legal processes.
5. Prioritize Digital Literacy and Education
Digital literacy education, including consent education, online safety, and awareness of deepfake technology, should be integrated into school curricula and public awareness campaigns. Prevention is as important as prosecution.
UN Women has warned that this is not a niche internet problem, but a “global crisis.”
- In a recent high-profile case, UK journalist Daisy Dixon discovered AI-generated, sexualized images of herself on X in December 2025, created using the platform’s own Grok AI tool. It took days for the platform to geoblock the function, while the abuse continued to spread.
- Deepfake abuse can serve as an online catalyst for so-called “honour-based crimes” in certain cultural contexts, where perceived breaches of honor norms on digital platforms can result in extreme physical violence against women, or even death.
- More than half of deepfake victims in the United States contemplated suicide, according to recent research.
Meanwhile, a handful of jurisdictions are starting to act:
- Brazil amended its criminal code in 2025, increasing the penalty for causing psychological violence against women using AI or other technology to alter their image or voice.
- The European Union Artificial Intelligence (AI) Act imposes transparency obligations around deepfakes.
- The United Kingdom’s Online Safety Act prohibits sharing digitally manipulated explicit images, but does not specifically address the creation of deepfakes and may not apply where intent to cause distress cannot be proven.
- The United States Take It Down Act explicitly covers AI-generated intimate imagery and requires platform removal within 48 hours.
The fight against deepfake abuse is far from over. The EU AI Act, for example, is still being implemented and its effectiveness remains to be seen. Ongoing monitoring of these legislative efforts, and continued advocacy for stronger protections, will be critical. The next key development to watch is the outcome of several pending legal challenges to platform immunity laws, which could significantly alter the landscape of online accountability.
This is a complex issue with no easy solutions, but inaction is not an option. The safety and well-being of women and girls depend on a swift and decisive response. Share your thoughts and experiences in the comments below.
Related reading
