Researchers Find 175,000 Publicly Exposed Ollama AI Servers Across 130 Countries
Researchers Find 175,000 Publicly Exposed Ollama AI Servers Across 130 Countries
https://thehackernews.com/2026/01/researchers-find-175000-publicly.html
Publish Date: 2026-01-29 13:37:00
Source Domain: thehackernews.com
- A significant number of unmanaged public AI compute infrastructures have been identified, predominantly based in China but also found across 130 countries.
- The exposed systems enable code execution, API interaction, and external system access, raising security concerns.
- Over 48% of the observed systems support tool-calling functions, expanding functionalities beyond text generation to potentially dangerous privileged operations.
- The widespread exposure of these systems can lead to “LLMjacking,” where bad actors abuse the infrastructure for activities like spam, disinformation, and cryptocurrency mining.
- The decentralized and often residential nature of the Ollama deployments creates governance gaps and new avenues for malicious activities.
- Researchers warn traditional security measures are insufficient, urging new approaches to protect unmanaged AI compute infrastructure similar to other exposed systems.
