Attackers are exploiting the buzz around generative AI (GenAI) to target less technical developers who may download free open-source Python packages without proper vetting. Two such packages, claiming to offer API access to popular chatbots like OpenAI’s ChatGPT and Anthropic’s Claude, were actually designed to deliver the "JarkaStealer" infostealer malware. These packages, "gptplus" and "claudeai-eng," were uploaded to the Python Package Index (PyPI) in 2023 under the user "Xeroline." They falsely advertised API integration but only offered basic interaction with ChatGPT’s demo version, while secretly deploying malware. JarkaStealer, sold on the Russian-language Dark Web for $20, is a simple but dangerous infostealer. It can capture sensitive data such as browser information, screenshots, and session tokens from apps like Telegram, Discord, and Steam. Though available on GitHub, its use is still risky, and its effectiveness in stealing data varies. These malicious packages remained on PyPI for about a year, managing to deceive developers into downloading them over 1,700 times across multiple operating systems in over 30 countries, including the U.S. The packages’ popularity seemed artificially boosted, particularly for "gptplus," which saw a sudden drop in downloads after the first day. This suggests attackers may have manipulated download statistics to create a sense of legitimacy. Security experts warn developers to verify a package's credibility before downloading and installing it, as many users, especially those less cautious, may fall victim to these types of attacks. Recently discovered and reported by Kaspersky researchers, the malicious packages were taken offline but serve as a reminder of the dangers of downloading unverified software.
Kaspersky researchers have uncovered a sophisticated supply chain attack targeting the npm ecosystem, where threat actors distributed the AdaptixC2 post-exploitation framework thro...
A critical vulnerability has been identified in Apache Syncope, allowing authenticated administrators to execute arbitrary code on affected systems. Tracked as CVE-2025-57738, this...
A critical vulnerability has been uncovered in the on-premise version of LANSCOPE Endpoint Manager, posing a serious security risk. Identified as CVE-2025-61932, the flaw exists in...