In Brief
- Digital technologies and generative AI are changing how violence and abuse are perpetrated, allowing harm to spread faster.
- WashU violence researchers say that many of the underlying dynamics — including power, prior exposure and social norms — are familiar.
- The new Center on Violence and Health is working to better understand emerging harms and develop prevention strategies.
Over the past two decades, digital technology has reshaped how violence is experienced and perpetrated. Social media, smartphones and other digital tools allow harassment, coercion, exploitation and abuse to move across settings, persist over time and reach audiences far beyond the original encounter.
Now, generative artificial intelligence (AI) is accelerating that evolution. Images can be fabricated with startling realism. Impersonation is becoming easier and more convincing. Abusive content can be created and distributed at unprecedented speed and scale. These technologies are changing the tools available to perpetrators — and raising new questions about how to understand, address and prevent violence.
Those questions are a growing focus for researchers at the WashU Bursky School of Public Health. In a recent Think Global Health commentary, Ilana Seff, Julia Fleckman and Lindsay Stark argue that holding technology companies accountable is only part of the response. Reducing exposure to potentially harmful digital environments, they write, is not the same as preventing violence or helping young people when violence occurs.
The three researchers also lead WashU’s new Center on Violence and Health. Stark, a professor, and Fleckman, an associate professor, serve as co-directors; Seff, a research associate professor, is associate director.
The interdisciplinary initiative housed in Bursky Public Health connects researchers studying forms of violence often treated as separate problems, including gender-based and firearm violence, trafficking, suicide, political violence and digital abuse.
Technology-facilitated violence and digital solutions is one of the center’s seven interconnected research programs. The center’s first symposium, planned for February, will examine gendered violence in digital spaces, including technology-facilitated abuse, online exploitation and deepfakes.
Here, Stark, Fleckman and Seff discuss what technology is changing about violence and what prevention research can teach us.
What is technology-facilitated violence, and what do we miss by thinking of it as simply online harm?
Seff: Technology-facilitated violence is any violence or abuse that happens through or is made worse by technology. It can include cyberbullying, harassment, threats, stalking, sharing intimate images and many other behaviors. But what happens online doesn’t stay online. Technology can make violence more persistent, expand the audience, give perpetrators anonymity, and leave people feeling like they’ve completely lost control over where harmful content goes. The use of Snapchat to facilitate sexual assault offline in the recent Cornell case serves as a prime example.
Are digital technologies creating new forms of violence, or changing the reach and consequences of existing violence? Where are new forms of harm emerging?
Seff: It’s really both. Technology can amplify forms of violence that already existed, like harassment, stalking and coercive control, but it’s also creating harms that weren’t possible before. Deepfakes are a good example. So is the ability to share intimate content instantly with thousands of people. The technology is evolving faster than our understanding of many of these newer forms of abuse.
How does technology-facilitated violence intersect with gender-based violence? Has technology changed who can perpetrate abuse or what power and control look like in a relationship?
Seff: Technology has given people new tools to carry out some very familiar forms of gender-based violence. A partner can monitor your location, control who you talk to, threaten to share intimate images or create fake sexual images of you. Technology can expand the reach and persistence of that abuse, but the underlying dynamics are familiar: power, control and exploitation.
How does AI change the problem?
Seff: This is where it gets scary. AI dramatically lowers the barrier to causing harm. Someone no longer needs an actual intimate image of you to create one, and abusive content can be generated and distributed almost instantly and at enormous scale. It also makes impersonation much easier, creating forms of abuse that simply weren’t possible in the same way a few years ago.
Your commentary in Think Global Health argues that reducing young people’s exposure to social media isn’t the same as reducing harm. What’s the distinction? And what do we misunderstand about the relationship between social media, violence and youth mental health?
Seff: It’s not necessarily that reducing young people’s time on social media doesn’t reduce harm. It’s that reducing time online doesn’t mean they won’t experience violence when they are online. Even if we reduce the time adolescents spend online, they will still be plugged into the digital ecosystem in one way or another, so we also need strategies aimed specifically at preventing violence and responding when it occurs. We’ve spent a lot of time asking whether social media itself makes adolescents anxious or depressed, but sometimes we skip over a more fundamental question: What is actually happening to them there? Harassment, threats, coercion, stalking and image-based abuse can all affect mental health. So the response can’t just be about screen time.
If a young person tells a parent, teacher or healthcare provider that they are being harmed online, what responses can help — and what can make the situation worse?
Seff: The first response matters a lot. Young people need to feel believed, not blamed or punished. Taking away their phone, minimizing the harm because it happened online, or asking why they shared an image can make them less likely to seek help again. Adults should focus first on safety, preserve evidence when appropriate, and ask what the young person wants and needs before taking action. Healthcare providers also need to be better equipped to recognize when online victimization may be affecting a young person’s mental health.
Are the ways we measure violence keeping pace with these changes? What might conventional surveys, crime statistics or other data sources fail to capture?
Seff: Not really. A lot of the existing measurement still focuses on cyberbullying and often reduces exposure to a yes-or-no question. That misses a huge amount of what shapes the experience and its consequences, including what happened, how long it lasted, who did it, whether threats were involved, how widely something was shared and whether it could be taken down. We’re currently reviewing the existing measures in this space, and one of our next steps at WashU is to pilot more comprehensive approaches that capture those dimensions. If we don’t measure the violence accurately, we can’t fully understand its effects — or know which prevention and response strategies actually work.
What have decades of research on other forms of violence taught us that could be applied to technology-facilitated violence?
Fleckman: We know a great deal about what contributes to violence and what can help prevent it. One important lesson is that exposure matters: People who experience or witness violence, or see it within their social networks, can be more likely to commit violence themselves. We also know that social norms and environments can either reinforce violence or help prevent it.
Those lessons apply online, too. Technology may change where and how violence occurs, but many of the underlying dynamics are familiar. That gives us a foundation for understanding how harmful behavior develops, when it may escalate and where we can intervene to prevent it.
The center brings together researchers studying forms of violence often treated separately. What becomes visible when you study them together?
Stark: The center’s seven programs — firearm injury, gender-based violence, health systems, conflict and political violence, child and adolescent well-being, mental health and substance use, and technology-facilitated violence — share overlapping risk and protective factors well established in the broader violence research field: prior exposure to violence; adverse childhood experiences; disrupted social support; and social norms that legitimize violence as an acceptable response.
Studying these forms of violence together rather than in separate silos allows us to see how violence in one domain can predict or compound risk in another. Firearm injury and intimate partner violence share risk markers. Conflict and displacement elevate rates of interpersonal and gender-based violence. Childhood adversity is a common thread running through nearly every category. Bringing these programs together allows us to study those connections — and translate that evidence into population-level programs and policies that address violence across traditional boundaries.
What aspect of technology-facilitated violence concerns you that researchers, policymakers or the public have not caught up with yet?
Stark: Generative AI is moving faster than our systems for recognizing or responding to it. Deepfake images, AI companion chatbots that can groom or manipulate, and synthetic sexual content are increasing, and most of our frameworks weren’t built for this. Image-based abuse laws often assume a real photo existed. Digital literacy programs teach kids to spot manipulation, but AI-generated content is increasingly difficult to distinguish from authentic content.
Schools, healthcare providers and even parents don’t yet have the language or training to recognize this as it’s happening, let alone respond to it. We are trying to catch up to problems from a few years ago while the technology continues to race ahead. That gap — between how fast the technology is changing and how slowly our response systems are moving — is deeply concerning. This is not hyperbole. Technology-facilitated violence is a critical public health issue, with documented consequences for morbidity and mortality, and it requires urgent research and intervention.
