Cybersecurity experts have identified a significant threat within the JetBrains Marketplace, where a ‘coordinated malware campaign’ has seen the release of 15 malicious plugins. These plugins, masquerading as AI coding assistants, are designed to steal API keys from AI providers.
Malicious Plugins Target AI API Keys
According to Ilyas Makari, a researcher at Aikido Security, each plugin is promoted as an AI coding assistant utilizing technology like DeepSeek. These plugins promise features such as code reviews and bug detection but are secretly exfiltrating the API keys entered by users to an attacker-controlled server.
The campaign, which has been active since late October 2025, includes plugins such as CodeGPT AI Assistant and DeepSeek AI Assist, each downloaded over 25,000 times. However, the authenticity of these download figures remains questionable.
Plugins Utilize Stolen API Keys
All identified plugins share a common codebase. Users are prompted to input an API key for services like OpenAI or DeepSeek to activate features. While functioning as advertised, the plugins send the API keys to a remote server at ‘39.107.60[.]51’ in plaintext, compromising user security.
Further analysis reveals a peculiar monetization strategy where users who pay a fee via the plugin’s donation system receive an API key from the server, which the plugin then uses for AI model interactions. This suggests that the attackers may be distributing stolen keys, offering unauthorized access to AI services.
Chrome Extensions Harvest AI Conversations
In a related development, two Google Chrome extensions, Smart Adblocker and Adblock for Browser, have been discovered capturing user interactions with AI chatbots. This data collection, termed ‘PromptSnatcher,’ records conversations and metadata from platforms like ChatGPT and Microsoft Copilot.
These extensions, operational since 2022 and 2023 respectively, utilize legitimate ad-blocking functions as a cover for their data interception activities. Despite this, it remains unclear if these practices breach Google’s extension policies.
Implications and Cautionary Measures
The exposure of these malicious activities highlights the growing risk of targeting developer environments and AI interactions. As these attacks become more sophisticated, developers and users should be vigilant when using plugins and browser extensions, especially those requiring sensitive information.
Security researchers advise treating plugins with the same caution as any dependency with access to sensitive data. Verifying the legitimacy of tools before inputting confidential information is crucial to maintaining cybersecurity.
As threat actors continue to exploit open-source and AI ecosystems, maintaining robust security practices and staying informed about such threats is essential for safeguarding personal and business data.
