Spy AI Microsoft has trained a massive new generative AI model — and strikingly, it’s entirely disconnected from the internet. The idea is to provide US intelligence agencies with a way to analyze top-secret information without running the risk of leaking any of it. As Bloomberg reports, the company claims it’s the first large language model (LLM) of its kind that’s completely divorced from the internet, unlike other competing ones that greatly rely on up-to-date information scoured from various places online. Microsoft’s new “air-gapped” AI, which is based on OpenAI’s GPT-4 LLM and will reportedly be available to 10,000 users, could allow agencies like the CIA to quickly analyze a wealth of classified data. In other words, civilians will likely never get their hands on it — or at least, Microsoft will do its best that won’t happen. “This is the first time we’ve ever had an isolated version — when isolated means it’s not connected to the internet — and it’s on a special network that’s only accessible by the US government,” Microsoft’s chief technology officer for strategic missions and technology William Chappell told Bloomberg. For CIA’s Eyes Only The conversation surrounding the unintentional dispersal of sensitive data via AI chatbots is as old as the tech itself. Executives of some of the largest companies out there have long warned their employees about making use of the tech, citing risks of leaking sensitive or proprietary information. Of course, the need for maintaining confidentiality becomes even higher when it comes…Microsoft Deploys Powerful New AI Completely Disconnected From the Internet