Emerging Threat: New Aoki Infostealer Versions Target AI Credentials
Recent findings from cybersecurity firm Symantec indicate a significant evolution in the capabilities of the Aoki infostealer, which now targets credential and session files related to popular AI coding assistants. This marks a notable shift from the earlier iterations of the Aoki malware that primarily sought out information within browsers, cryptocurrency wallets, and password managers, leaving AI-related assets untouched.
The updated version of Aoki has expanded its reach to include API keys and tokens from well-known AI coding assistants such as Anthropic’s Claude, OpenAI’s Codex, SpaceX’s Cursor, and Google’s Gemini. By gaining access to these credentials, malicious actors can engage in nefarious activities while masquerading as legitimate developers. Furthermore, this can circumvent the security measures that typically assume requests originate from trusted users.
The implications of such breaches are profound, as the compromise of these tokens can expose connected repositories, cloud services, and continuous integration/continuous delivery (CI/CD) pipelines associated with the victim’s accounts. These risks elevate the potential for widespread disruption in development environments that rely heavily on these tools.
Aoki’s Focus on AI Credentials
According to the report released on October 15, the Aoki infostealer is a relatively new entity in the cybersecurity landscape, first identified in early 2026. The ongoing development of its capabilities is evident in its recent targeting of files specific to AI systems, such as .claude.json, Claude desktop cookies, ChatGPT cookies, as well as the Codex configuration files, config.toml, and .env.
Moreover, Aoki now also siphons off tokens for users of GitHub Copilot, Cursor, Gemini, JetBrains AI Assistant, and tools like Kiro, as well as Amazon Web Services (AWS) credentials. These advancements lead researchers to classify Aoki as a highly adept and multi-functional infostealer, with capabilities still undergoing active enhancement.
Rising Value of AI Credentials
Symantec experts have observed that the development of Aoki reflects a growing trend where AI-related credentials are now considered as valuable to hackers as traditional financial data, such as payment card information. Much like card data, a stolen AI key can lead to direct monetary gain, either through resale in underground markets or by exhausting the victim’s account quotas in a practice commonly referred to as "LLMjacking." What’s particularly alarming is the added potential for attackers to infiltrate not just individual accounts, but broader development environments.
The ongoing evaluation of this threat landscape has revealed that the motivations behind these attacks are likely financially driven. Despite the sophisticated nature of Aoki and its apparent evolution, Symantec remains uncertain about the specific threat groups that may be orchestrating these attacks, as there are currently no established links to any existing malware families.
The implications for developers and organizations relying on AI tools are alarming. As AI technology continues to flourish and become deeply integrated into various sectors, the risk posed by such stealer malware also multiplies. Users must bolster their security measures, implementing strategies like multifactor authentication and monitoring for unusual account activity, to safeguard against potential breaches.
Conclusion
The emergence of Aoki infostealer versions that specifically target AI credentials underscores the shifting landscape of cyber threats. Companies and developers must remain vigilant as cybercriminals adapt to capitalize on the security vulnerabilities associated with rapidly evolving AI technologies. The necessity for rigorous security protocols and awareness amidst this evolving threat cannot be overstated—everyone from individual developers to large organizations must prioritize cybersecurity measures to protect sensitive information from exploitation.
As the reach and capabilities of threats like Aoki expand, proactive risk management strategies become increasingly vital to defend against this new breed of cybercrime focused on exploiting both human and technological shortcomings in the digital age.