Canto Incognito Malware Targets Exposed AI and LLM Infrastructure, Infects 3,400+ Servers

A new malware family operating under the name Canto Incognito has been observed targeting exposed artificial intelligence and large language model infrastructure, according to cybersecurity researchers. The financially motivated campaign has infected more than 3,400 servers worldwide, with the primary objective of deploying cryptocurrency miners and expanding the botnet's scale.

The threat actors behind Canto Incognito have been identified as exploiting exposed AI and LLM infrastructure to gain initial access. Once compromised, the malware installs cryptocurrency mining software, leveraging the victim's computing resources for financial gain. The infection of over 3,400 servers represents a significant expansion of the botnet's capacity.

Reporting indicates that the campaign specifically targets systems where AI and LLM endpoints are exposed to the network without adequate security controls. The exact initial access vector details have not been fully disclosed in available reports, but the exploitation of exposed infrastructure is the common theme. The deployment of cryptocurrency miners on compromised systems allows the threat actors to monetize the botnet through illicit mining operations.

The emergence of Canto Incognito highlights the growing risk associated with the rapid deployment of AI infrastructure, particularly when security best practices such as network segmentation and access controls are overlooked. Security researchers recommend that organizations review the exposure of their AI and LLM systems and implement appropriate defensive measures.