Can Chatbots Help Prevent Intimate Partner Violence?

Generative AI is rapidly reshaping our intimate lives, for better and for worse. On one hand, people are increasingly turning to AI chatbots for companionship, relationship advice, and emotional support. On the other, AI-powered image generation has fueled a surge in deepfake pornography and other forms of nonconsensual intimate image abuse.

These contrasting developments raise an important question: Can the same technology that creates new risks also be used to reduce harm? In this post, we’ll explore new research examining whether AI chatbots can provide accurate, supportive information about intimate partner violence (IPV) and sexual violence. The findings offer insight into whether generative AI has the potential to become a meaningful tool for prevention and intervention, despite its well-documented shortcomings in other areas.

Intimate Partner Violence and Sexual Violence 

Intimate partner violence refers to harm that occurs between romantic or sexual partners, such as domestic abuse, emotional or financial abuse, and sexual abuse. IPV encompasses all forms of violence and aggression toward a current or previous partner, and it can occur both in-person and in technologically-mediated ways (e.g., cyber stalking). IPV affects over 1 in 3 women and 1 in 6 men in the United States over the course of their lifetimes, and it carries long-term health consequences for victims, both emotionally and physically.

Experiencing IPV can be incredibly isolating. A commonly-referred to statistic is that it takes survivors an average of seven attempts to leave an abuser. This is oftentimes due to the multifaceted ways that perpetrators of IPV establish power and control over their partners. Contrary to popular belief, leaving an abusive relationship is not always the safest or most realistic option. In some situations, attempting to leave increases the risk of violence. In others, an abuser’s control over finances, housing, transportation, or other essential resources makes leaving nearly impossible.

Can AI Help Mitigate IPV? 

Lack of knowledge about resources supporting victims of IPV is a large hurdle to mitigating it. In an age of information accessibility via the internet and GenAI, chatbots may serve as a way for people to learn more about possible signs of IPV, as well as how and where to get help. 

Of course, there are important concerns about using AI chatbots to learn about IPV, with privacy being perhaps the most significant. For someone experiencing abuse, even seeking help can be dangerous. Questions about whether a partner can access chatbot conversations, whether a user’s identity can be inferred from their interactions, and how those conversations are stored or used over time all underscore the potential risks that AI may pose. [1] Chatbots may also do more harm than good if they provide stigmatizing, dismissive, or inaccurate responses about IPV, potentially discouraging users from seeking additional help or support.

A recent study published in The Journal of Sex Research underwent a scoping review of current research that investigated AI chatbots’ performance in IPV contexts. [2] Specifically, studies had to “focus on (1) how AI chatbots serve as a tool for preventing intimate partner or sexual violence or (2) how AI chatbots respond to questions on intimate partner or sexual violence” (p. 4). Based on this criteria, 40 articles, pre-prints, and conference papers were reviewed by the research team. 

The researchers found that chatbots were largely developed to provide tertiary support when it came to IPV, which serves to mitigate harm and consequences after violence has occurred. Chatbots were also able to answer questions about sexual violence writ large, but had a much more difficult time parsing out details of interactions when presented with a vignette scenario. Fewer chatbots were developed to provide primary or secondary support, which are more focused on preventing violence from occurring in the first place. These other forms of support focus on self-advocacy skills and bystander intervention, as well as how to recognize the signs of IPV. 

Overall, participants preferred chatbot responses that were warm, empathetic, and supportive—more like talking to a trusted friend or counselor than interacting with a search engine (though they did not view AI as a replacement for therapy). As the authors note, “AI was generally envisioned as a supplement to human care, not a substitute” (p. 18). At the same time, participants saw chatbots as particularly useful for practical tasks, such as helping them navigate the legal system, develop a safety plan, or understand the process of reporting abuse.

Thinking About The Future Of AI and IPV

As research in this area continues, privacy and ethics must remain at the forefront. However, given the chronic underfunding of IPV services and the limited availability of trained advocates, AI chatbots could eventually serve as an important first point of contact for people seeking help. In fact, the National Domestic Violence Hotline has already introduced an AI chatbot that users can access when a live advocate is unavailable, illustrating how AI may help relieve pressure on overstretched support systems.

Looking ahead, chatbots appear especially promising as a way to connect people with reliable, evidence-based information and resources, much like a well-curated Google search. The scoping review found that advanced AI models generally performed well at identifying signs of IPV and directing users to appropriate support. However, they were far less capable of recognizing the subtle relationship dynamics that often characterize abuse or interpreting the nuanced ways people describe their experiences. [2] In other words, while AI may be effective at helping users find resources or begin developing a safety plan, it still struggles with identifying abuse when it is described indirectly or ambiguously.

To reach their full potential, future chatbots will need to be trauma-informed, culturally responsive, privacy-conscious, and resistant to generating misinformation. Additional research is also needed to understand how people experiencing IPV naturally communicate with AI, how chatbots respond in real-world interactions, and where these systems succeed—or fall short—in providing meaningful support.

If you or someone you know is experiencing IPV, visit https://www.thehotline.org/ to be connected with resources, chat with a representative, and make a safety plan. 

If you have a sex question of your own, record a voicemail at speakpipe.com/sexandpsychology to have it answered on the blog or the podcast. 

Want to learn more about Sex and Psychology? Click here for more from the blog or here to listen to the podcast. Follow Sex and Psychology on Facebook, Twitter (@JustinLehmiller), Bluesky, or Reddit to receive updates. You can also follow Dr. Lehmiller on YouTube and Instagram.

Image made in Canva.

References 

[1] Butterby, K., & Lombard, N. (2025). Developing a chatbot to support victim-survivors who are subjected to domestic abuse: Considerations and ethical dilemmas. Journal of Gender-Based Violence, 9(1), 153–161. https://doi.org/10.1332/23986808Y2024D000000038

[2] Marcantonio, T. L., Vowels, L. M., LeRoux, S., Bravo, M., Thrash, A., Mueller, A., Niolet, A., Hall, M., Döring, N., & Nadarzynski, T. (2026). The Role of Artificial Intelligence in the Prevention of Intimate Partner Violence and Sexual Violence: A Scoping Review of the Literature. The Journal of Sex Research, 1–25. Advance online publication.  https://doi.org/10.1080/00224499.2026.2686406