US Cyber Command wants to run AI models from OpenAI, Google, and others on the most sensitive systems used by US intelligence agencies. The push comes as AI with advanced hacking capabilities develops at a rapid pace.
US Cyber Command has launched a new task force to speed up deployment of powerful AI models on highly classified Pentagon and NSA networks. That's according to Politico, citing three people familiar with the effort.
General Joshua Rudd, who leads both the NSA and Cyber Command, announced the task force two weeks ago in an internal email. It spans both organizations and will assess how AI models from Silicon Valley companies can be safely deployed on so-called "high-side" systems - networks with the highest classification levels. Technical expertise will come mainly from the NSA's AI Security Center, while a Cyber Command officer will lead the effort.
The immediate trigger, Politico reports, is the development of AI models that can find security vulnerabilities in digital systems faster than the best human hackers. In April, Anthropic released its Claude Mythos model and restricted access because the consequences of misuse could be severe for the economy and national security. OpenAI has since announced models with similar capabilities. Anthropic estimates comparable tools could be widely available within six to 24 months.
Lt. Gen. Charles Moore, former deputy commander of Cyber Command, told Politico the task force isn't just a good idea but a necessity. He said AI tools are becoming essential for threat detection, vulnerability prioritization, and faster decision-making in both offensive and defensive cyber operations.
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。