Amazon Uncovers AI Agent Campaign Targeting LiteLLM, Dify, and Other Apps for API Keys
AI Image

Amazon Uncovers AI Agent Campaign Targeting LiteLLM, Dify, and Other Apps for API Keys

The Hacker News • • business

Key Points:

  • Attackers using AI agents, notably the ARTEX tool, have been targeting internet-exposed AI applications like LiteLLM, One API, and Dify to steal AI provider keys, which they then use to further search for additional keys and exploit vulnerabilities.
  • Amazon's threat intelligence team captured records from decoy servers showing these AI agents exploiting default credentials, unfinished installs, and known security flaws, with moderate confidence linking the operators to China-based groups.
  • Cloudflare reported incidents where ARTEX accessed compromised API tokens without restrictions, leading to extensive unauthorized database requests and attempts to establish persistence within customer environments, often facilitated by static CI/CD tokens.
  • The FBI issued an alert highlighting widespread use of ARTEX since August across multiple sectors globally, noting the increased speed and persistence AI agents bring to attacks, while ARTEX's official GitHub was taken down but forks and commercial versions remain available.
  • Security experts recommend treating AI apps and inference endpoints like critical production systems by changing default credentials, scanning for exposed secrets, enforcing token restrictions, monitoring for suspicious activity, and applying patches for known vulnerabilities such as those in Langflow and RAGFlow.

Trending Business

Trending Technology

Trending Health