A new Windows infostealer malware called Volta-Autodoxer has been identified, which exfiltrates sensitive user data including geolocation, employment details, partial credit card numbers, and seed phrases. This stolen information is then sent to large language models via OpenRouter to assess the monetization potential of victims. The malware employs a chain of models, including Claude Opus 4.6, Kimi K2, and Gemini 2.5 Flash, with integrated refusal detection and data exfiltration over Telegram, according to Talos analysis. AI
IMPACT This malware highlights the potential for LLMs to be misused in cybercrime for victim assessment and monetization, posing new security challenges.
RANK_REASON Discovery of new malware that uses LLMs as a component.
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →