Pentagon's Use of Agentic AI Raises Cybersecurity Concerns Amidst Nation-State Threats
The Pentagon has been utilizing agentic AI tools on its GenAI.mil platform to significantly reduce the time required for various tasks, compressing weeks of work into mere hours. According to Emil Michael, the defense undersecretary for research and engineering, this initiative has been a 'tremendous success' since its rollout in December. However, these same tools are also being exploited by cybercriminals, granting them capabilities akin to nation-state actors. The Pentagon is currently in a testing phase with Mythos, an agentic AI model developed by Anthropic, a company labeled as a national-security risk. Despite this designation, the model is being used to identify vulnerabilities in both government and private sector infrastructures, including rural hospitals and wastewater treatment plants. 
