AI Doxxing Tools: Extremist Forums Automate Harassment
Far-right internet communities have begun transforming doxxing into a nearly automated process. Users of the anonymous imageboard Soyjak Party engineered several new applications. These instruments actively hunt for data leaks. Furthermore, they cross-reference information from public databases. Consequently, they compile comprehensive dossiers on targeted individuals in mere minutes.
Open Measures researchers discovered discussions surrounding at least three such services. These specific tools have emerged since October 2025. You can read more about these Soyjak AI doxxing tools in their recent report. The developers avoided training a dedicated model from scratch. Instead, they connected existing language models to third-party breach databases and search interfaces.
The Rapid Mechanics of Automated Targeting
First, a user inputs a target name or other initial data. Then, the system aggressively hunts for email addresses and user accounts. Additionally, it targets IP addresses, geolocation coordinates, and other identifiable information. The initially discovered tool supported bulk searching and source filtering. It also featured IP-based geolocation tracking to map targets accurately.
The creator distributed this service among forum members. He actively requested bug reports and feature suggestions. Subsequently, malicious users tested the program against real human targets. Meanwhile, the developer incrementally integrated their requested functions.
Autonomous Agents and Hallucination Risks
In March 2026, another dangerous service surfaced on the Soyjak Party imageboard. The author proudly labeled this creation a fully autonomous AI doxxing agent. A hidden system prompt forced the language model to roleplay as a private investigator. This persona relentlessly hunted for personal identifying details. The creator boldly claimed the service could generate a complete dossier from a single prompt. Ultimately, this process required almost zero operator intervention.
Open Measures uncovered completed intelligence reports targeting at least four private individuals on the imageboard. Furthermore, researchers found mentions of these tools on the Kiwi Farms forum. They also spotted them within a Discord channel. Therefore, this dangerous technology has already escaped the confines of Soyjak Party. To prevent platform proliferation, the investigators deliberately concealed the service names and website addresses.
Lowering the Barrier to Digital Harassment
Members utilize the main Soyjak Party section to coordinate harassment, issue threats, and expose private data. Participants frequently target minority representatives, public activists, and content creators. Consequently, automation drastically lowers the barrier to entry. It empowers individuals lacking basic investigative skills to harvest sensitive intelligence easily.
However, artificial intelligence cannot guarantee absolute data accuracy. The forum users themselves warned others about language model hallucinations. These glitches can mistakenly fuse details concerning entirely different people. Alternatively, they might inject completely fabricated elements into the files.
Nevertheless, an erroneous dossier does not mitigate the underlying threat. Fabricated data can easily trigger a vicious harassment campaign against an innocent bystander. Conversely, accurate physical addresses and phone numbers severely elevate real-world dangers. These physical threats include stalking, death threats, and malicious swatting incidents. Ultimately, researchers consider the emergence of these services a terrifying new phase in online harassment.
Support Our Threat Intelligence
If you find our technology report and cybersecurity news helpful, consider supporting our work.
Buy Me a Coffee
PayPal
