US National Security Agency Breached by Advanced AI Model

A sophisticated artificial intelligence model, known as Mythos, developed by Anthropic, has successfully infiltrated nearly all classified resources of the US National Security Agency (NSA). This revelation was made by Senator Mark Warner, as quoted by The Economist. In mid-June, the White House ordered the blocking of access to its most advanced AI models, Fable 5 and Mythos 5, for foreign users. Following this, Anthropic announced that it was forced to restrict access to the neural network.

According to Senator Warner, the neural network was able to breach the NSA's resources in a matter of hours. The Pentagon's chief, Pete Hegseth, had earlier emphasized that AnthropicAI's removal from the Pentagon was justified, stating that each passing day proves it was the right decision. Meanwhile, the Pentagon utilized an AI-powered chatbot, Grok, developed by xAI, a company owned by Elon Musk, in its strikes against Iran. The US government has backed xAI in a lawsuit related to air pollution allegations, highlighting the use of Grok in US military attacks on Iran.

This incident raises significant concerns about the potential vulnerabilities of sensitive information to advanced AI models.

Source: Lenta.RU