GCD AI RADAR
incidents/AIR-2025-0056

Threat Actors Hijack Cloud-Hosted LLMs via Stolen Credentials

Attack on AI3 Confirmed exploitationOtherData leak

Malicious actors utilized stolen credentials to access cloud-hosted large language models and created a reverse proxy for illegal resale. The activity impacted various cloud services and AI providers.

Details

type
Attack on AI
An AI model, agent or AI-powered product was the thing exploited, abused or leaked.
categories
  • Otherprimary
    In scope, but none of the categories above fits. Reviewed periodically for new categories.
  • Data leak
    Training data, chat logs, system prompts, model weights or other AI-system data exposed.
owasp llm top 10
mitre atlas
no technique for these categories
Both rows are derived from the type and categories (OWASP 2025 edition, ATLAS release 2026.09), not classified per record. How
severity
3 · Confirmed exploitation
Used against real targets; limited or unclear impact.
targets
OpenAI, Anthropic, Mistral, Amazon, Google, Framework / tooling
products
AWS Bedrock, Azure AI, GCP Vertex AI
incident date
6 May 2024
first reported
22 Apr 2025
origin
MITRE ATLAS
Derived from MITRE ATLAS (Apache 2.0, © MITRE).
classifier confidence
0.95 · published at or above 0.80; see the methodology
tags
llm-jacking · credential-theft · reverse-proxy
record
created 2026-10-18 06:55Z · updated 2026-10-18 06:55Z · json

Sources · 5

  1. 2025-04-22
    LLM Jacking
    MITRE ATLAS · atlas.mitre.org/studies/AML.CS0030
  2. 2025-04-22
    LLMjacking: Stolen Cloud Credentials Used in New AI Attack
    sysdig.com · sysdig.com/blog/llmjacking-stolen-cloud-credentials-used-in-new-ai-attack
  3. 2025-04-22
    The Growing Dangers of LLMjacking: Evolving Tactics and Evading Sanctions
    sysdig.com · sysdig.com/blog/growing-dangers-of-llmjacking
  4. 2025-04-22
    LLMjacking targets DeepSeek
    sysdig.com · sysdig.com/blog/llmjacking-targets-deepseek
  5. 2025-04-22
    AIID Incident 898: Alleged LLMjacking Targets AI Cloud Services with Stolen Credentials
    AI Incident Database · incidentdatabase.ai/cite/898

Titles and links belong to their publishers. The summary above is written by GCD AI RADAR in its own words; data is CC BY-SA 4.0.

Related incidents