Researchers at security firm MalExt Sentry have uncovered a malicious campaign targeting users of modern AI services. Two extensions exploited the trust of people who simply wanted a cleaner web browsing experience without annoying ads. While the programs actually blocked ad content using publicly available filters, they were carrying out extensive spying in the background. To hide their actions, they offered users the innocent option of opting in to \u201cenhanced protection,\u201d which made no mention of collecting chat data.
The first and significantly more widespread application is called \u201cSmart Adblocker\u201d (with the ID code iojpcjjdfhlcbgjnpngcmaojmlokmeii), which was installed by about 80,000 people. The second abused extension is called \u201cAdblock for Browser\u201d (with the ID code jcbjcocinigpbgfpnhlpagidbmlngnnn), which had about 10,000 users. Both used the same backend infrastructure and a hidden data drive, which experts internally named \u201cPanel 231\u201d.
The embedded mechanism targeted eight established AI platforms, namely ChatGPT, Gemini, Claude, Copilot, Perplexity, DeepSeek, Grok and Meta AI. The tool directly connected to the data traffic of the aforementioned websites and transcribed entire conversations. The system was able to store up to 10,000 characters for user questions and up to 30,000 characters for AI answers. In addition to the content itself, the extensions also recorded data about the AI model used and information about whether the user had a paid subscription. All this data was then transferred to the malicious servers.
What's particularly insidious is that the Firefox versions explicitly claimed not to collect any data, while doing the exact opposite. Because people often enter very personal things into AI chats, from health and financial issues to work logins, passwords, and internal company data. This makes such abuses a huge risk.
If you have either of these extensions installed, remove them immediately. Instead, opt for established, open-source solutions for blocking ads.<\/p>","protected":false},"excerpt":{"rendered":"
Raziskovalci varnostnega podjetja MalExt Sentry so razkrili zlonamerno kampanjo, ki je ciljala na uporabnike sodobnih storitev umetne inteligence. Dve raz\u0161iritvi sta namre\u010d izkori\u0161\u010dali zaupanje ljudi, ki so \u017eeleli zgolj \u010distej\u0161e brskanje po spletu brez nadle\u017enih oglasov. \u010ceprav sta programa dejansko blokirala oglasne vsebine s pomo\u010djo javno dostopnih filtrov, sta v ozadju izvajala obse\u017eno vohunjenje. Za […]<\/p>","protected":false},"author":2,"featured_media":0,"comment_status":"","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[70],"tags":[478,482],"class_list":["post-7757","post","type-post","status-publish","format-standard","hentry","category-spletni-brskalniki","tag-informacijska-varnost","tag-kibernetska-varnost"],"acf":{"subtitle":"","heading":"","summary":"Raz\u0161iritvi Smart Adblocker in Adblock for Browser sta zlorabili zaupanje 90.000 uporabnikov, saj sta pod krinko blokiranja oglasov na skrivaj snemali in odtujevali njihove osebne ter poslovne pogovore z osmimi AI platformami.","thumbnail_small":"https:\/\/racunalniske-novice.com\/wp-content\/uploads\/2021\/06\/does-malware-exist-560x315.jpg","thumbnail_large":"https:\/\/racunalniske-novice.com\/wp-content\/uploads\/2021\/06\/does-malware-exist.jpg","thumbnail_caption":"Zlonamerna koda Vigilante vam bo prepri\u010dala dostop do najbolj znanih piratskih strani.","gallery":"","video_gallery":null,"author":"","links":null,"sources":[{"title":"notebookcheck","url":"https:\/\/www.notebookcheck.net\/Malicious-ad-blocker-extensions-caught-secretly-reading-users-AI-chats.1331698.0.html"}],"skip_language":[]},"yoast_head":"\n