Seeking Help in the Digital Age: A Cross-Platform Analysis of Online Support Systems for Technology-Facilitated Abuse Victims
This paper presents a large-scale cross-platform evaluation of online support systems for technology-facilitated abuse victims, revealing that while search engines and general-purpose AI offer more actionable guidance than peer forums, none consistently provide safe, trauma-informed support, with significant risks including malicious links, toxic responses, and inadequate safety-aware guidance across all channels.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine the internet as a massive, bustling city. In this city, there are libraries (search engines), town squares where people shout advice to anyone who will listen (social media forums), and friendly robots that promise to answer any question you ask (conversational AI). For most people, this city is a place to find recipes, learn math, or chat with friends. But for some, this city has become a trap. When someone is being stalked, harassed, or controlled through their phone or computer, they often run to this digital city for help, hoping to find a map out of the danger. This is the world of "technology-facilitated abuse," where the very tools we use to connect—like GPS trackers, messaging apps, and smart home devices—are turned into weapons. The big question researchers have been asking is: If a victim runs into this digital city to ask for help, will the city guide them to safety, or will it accidentally lead them into a darker alley?
A team of researchers decided to play detective in this digital city to find out. They didn't just ask the robots or read the signs; they built a massive simulation using thousands of real questions that victims had actually asked over the last ten years. They treated these questions like test drives, sending them into three different parts of the internet: Google Search, Reddit (a giant online forum), and various AI chatbots. They wanted to see if the answers were helpful, if they were safe, and if they were kind.
The results were a bit like a rollercoaster ride with some scary drops. First, they found that Google Search and the big, general-purpose AI robots were actually pretty good at finding the right type of answer. If you asked, "How do I stop someone from tracking my phone?" they usually found a webpage or a robot response that talked about phone tracking. However, being "on topic" wasn't the same as being "safe." The researchers discovered that more than 65% of the time, when victims clicked on the search results, they ran into dangerous traps—links to phishing scams or malware. It was as if the library gave them the right book, but the book was filled with poison.
On the town square of Reddit, the story was different. While people were often willing to talk, the advice was rarely useful. Less than half of the questions got a real answer, and when they did, the advice was often vague or incomplete. Worse, about 20% of the time, the people shouting advice were actually mean, toxic, or blaming the victim. It was like asking for directions in a town square and getting shouted at or told to "just walk faster" instead of being given a map.
The most surprising twist came from the robots. The researchers tested special chatbots designed specifically to help victims of abuse, expecting them to be the heroes of the story. Instead, the general-purpose AI robots (the ones that answer everything from "how to bake a cake" to "how to fix a computer") actually gave better technical advice than the specialized ones. The specialized bots often failed to answer the question at all or gave generic, unhelpful responses. But here is the catch: even the "better" general robots had a major flaw. While they were good at giving steps to fix a computer, they were often terrible at understanding the human fear behind the question. They frequently gave advice that was technically possible but dangerous for a victim, like telling someone to "just delete your account" without warning that this might destroy evidence needed to catch the abuser. They also rarely mentioned where to find a real human helper, like a hotline or a shelter.
In the end, the study suggests that while the digital city has plenty of information, it is not yet a safe place for someone in danger. The tools we have today—whether it's a search engine, a forum, or a specialized robot—are not good enough on their own. They might give you the right words, but they often miss the most important part: keeping you safe while you try to escape. The researchers conclude that we need to rebuild these digital helpers to be not just smart, but also careful, kind, and deeply aware of the risks, ensuring that the next time someone runs to the internet for help, they find a lifeline instead of a trap.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.