AI-Driven Doxing Tools Emerge in Online Forums, Warns Kaspersky
Kaspersky, a cybersecurity firm, has warned that artificial intelligence (AI) is enabling faster doxing—unauthorized collection and dissemination of personal information—by linking scattered digital data. The firm cited a report by Open Measures, which identified users on the forum Soyjak.party discussing AI tools to expose anonymous users as early as October 2025. Mobitekno reported that these tools target activists, bloggers, and forum participants perceived as adversaries.
AI Exploits Online Data to Reconstruct Identities
The AI systems in question automate the aggregation of open-source information, including email addresses, usernames, and metadata from photos and videos. For example, location data in image metadata or IP addresses can help pinpoint a user’s location. A February 2026 study referenced by Kaspersky analyzed 338 posts from Hacker News and found that AI correctly identified 226 individuals—about two-thirds of the sample—based on job histories, interests, and backgrounds. The cost to uncover identities ranged from $1 to $4 per person, highlighting the low barrier to entry for malicious actors.
Photos and videos also provide clues through metadata, such as timestamps, locations, and device models, which can reveal a user’s activities and environment. Kaspersky noted that AI systems can link connected email addresses and usernames to expand their search, creating detailed profiles from fragmented data.
Traditional OSINT Tools Lack AI’s Autonomy
Unlike AI-based systems, conventional Open Source Intelligence (OSINT) tools operate via pre-programmed scripts and lack the ability to “think” or make spontaneous decisions. Teknogav.com noted that while some AI-powered doxing tools on Soyjak.party are no longer active, similar technologies have spread to other image-based forums. The report emphasized that even casual forum conversations can inadvertently reveal identities, as seen in the Hacker News study.
Source 2 added that perangkat serupa (similar tools) have moved to other image-based forums, though those on Soyjak.party are now non-functional. The study also highlighted that everyday forum interactions, such as shared interests or routine posts, can expose personal details, as noted in a February 2026 analysis.
Kaspersky Recommends Steps to Minimize Digital Footprints
To mitigate risks, Kaspersky advised users to:
– Conduct regular online searches for their name, email, phone numbers, and nicknames to identify exposed accounts.
– Restrict social media profiles to private settings and delete outdated content.
– Avoid linking personal identities across services, using unique email addresses and usernames for different contexts.
– Enable two-factor authentication (2FA) with time-based one-time passwords (TOTP) via trusted password managers.
Source 1 emphasized using distinct email addresses and usernames for different contexts, avoiding connections between anonymous accounts and public profiles or easily identifiable recovery addresses. Source 2 also recommended using the free online Privacy Checker to adjust device and app privacy settings.
Privacy Protections Require Proactive Measures
The firm also recommended removing metadata from photos and videos before sharing them online. Users should monitor data breach notifications and update passwords promptly. Kaspersky’s Password Manager alerts users to reused passwords and compromised credentials, while its Premium service tracks linked accounts.
Source 2 added that Kaspersky’s security solutions can prevent online personal data collection and report existing leaks. Users should also close unused accounts, as they often retain personal data for years and pose a risk during breaches.
Challenges in Erasing Online Identities
Kaspersky acknowledged that completely disappearing from the internet is nearly impossible. Instead, users should focus on reducing “identity linkages”—connections that enable data reconstruction. This includes closing unused accounts, which often retain personal data for years, and regularly reviewing app and device privacy settings.
Source 1 noted that reducing identity linkages is critical, as even minor connections—like a forgotten email address or a shared username—can reconstruct a profile. The firm also warned against using AI to generate self-profiles, as this could inadvertently expose sensitive data.
Industry Response and Ongoing Risks
Source 2 highlighted that while AI-driven doxing remains a threat, proactive measures like two-factor authentication with TOTP and regular account audits can significantly lower risks.