Threat Actors Hijack Cloud-Hosted LLMs via Stolen Credentials

# Threat Actors Hijack Cloud-Hosted LLMs via Stolen Credentials

https://gcdradar.vercel.app/incidents/threat-actors-hijack-cloud-hosted-llms-via-stolen-credential · AIR-2025-0056 · closed 2025-05-06

- Lane: Patch now
- Severity: 3 (Confirmed exploitation)
- Fix status: Fix status unknown
- First reported: 2025-04-22 (happened 2024-05-06)
- AI role: Attack on AI
- Products: AWS Bedrock, Azure AI, GCP Vertex AI

## Summary

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.

## What to do

No action recorded; read the sources below and the vendor's advisory.

## Latest updates

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

## Sources

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

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.