Gbhackers Amazon Exposes AI Campaign Targeting API Keys with ARTEX Framework
Article Content
- •AI agents used ARTEX to steal API keys from exposed applications.
- •The campaign targeted multiple AI platforms, including LiteLLM and Dify.
- •Amazon captured operational records via its MadPot decoy network.
Amazon's threat intelligence team revealed a campaign where AI agents used the ARTEX framework to steal API keys from various AI applications, including LiteLLM and Dify. The attacks occurred between October 3 and November 5, 2026, with the agents targeting internet-exposed applications to extract provider keys. These stolen keys were then utilized to power further credential theft operations, creating a self-sustaining cycle. Amazon captured the agents' operational records through its MadPot decoy network, which posed as popular LLM APIs. The attackers, believed to be linked to multiple groups in China, used prompts in Simplified Chinese and avoided targeting mainland China. The campaign exploited known vulnerabilities, including CVE-2025-3248 and CVE-2026-28797, and focused on applications that store multiple provider keys, making them particularly vulnerable. Amazon has not confirmed the success of the attacks but assessed the threat with moderate confidence.
Ask AI about this cluster
Answers cite the sources they use
Timeline
More articles in this cluster (8)
Following this threat?
Track blacknevas, Amazon and CVE-2025-3248 in your own feed — you're alerted when they show up in new reporting, leak sites or exploitation.
Free account · no card needed
Common questions
Which applications are affected?
What vulnerabilities are involved?
How can we protect against this threat?
Continue Reading
Falcon C2 Framework Compromises Government Browsers in Southeast Asia and Middle East Unit 42 has identified the Falcon command-and-control framework, also known as XG or Xiàng Gē, which has been used to compromise browsers of government entities in one Middle Eastern country and four Southeast Asian countries between March and May 2026. The framework executes malicious JavaScript within the browser…
GhostFrame Malware Targets VMware ESXi Servers Amid Ransomware Attacks Recent investigations by Kaspersky and Insikt Group revealed that ransomware attacks on ESXi and Proxmox environments exploited weak administrative controls and misconfigurations rather than VM escape techniques. The Kaspersky Global Emergency Response Team linked these incidents to BlackNevas and The Gentlemen…