Artificial intelligence startup Anthropic has been revealed to have reached a deep collaboration with the U.S. National Security Agency (NSA), assisting it in deploying the latest high-performance Claude Mythos large model for offensive cyber operations. According to insiders, Anthropic even sent about six technical engineers to the NSA for on-site support, responsible for guiding the implementation of the model and conducting in-depth customization and optimization of the AI model according to specific operational needs.

Controversial Technology with Both Offensive and Defensive Capabilities

As a tool with significant security risks, the Claude Mythos large model was initially only available to a limited number of U.S. domestic institutions upon its release in April. However, this week it announced that it would expand its authorization to 150 institutions across 15 countries globally. The model is capable of autonomously discovering and exploiting software vulnerabilities, and its powerful dual capabilities of offense and defense have sparked widespread concerns among governments, financial institutions, and IT companies around the world.

In response to public concerns about the risk of technology abuse, an insider close to Anthropic defended the company, stating that mastering advanced attack technologies is the optimal way to build a strong defense system. They believe that adversaries are likely also developing similar AI weapons, and if their side does not actively develop them, enemies will find ways to implement similar technologies. Therefore, it is essential to seize the strategic advantage in geopolitics.

The Struggle Between Legal Disputes and Business Ambitions

Notably, this highly sensitive military cooperation took place as Anthropic was facing intense legal disputes with the U.S. Department of Defense over the use of AI technology in warfare. Previously, Anthropic was placed on the supply chain risk list by the military after refusing to use the Claude series models for large-scale surveillance and the development of lethal autonomous drones. If the lawsuit fails, the company may be forced to terminate all commercial contracts with military agencies.

Despite being involved in legal disputes and facing compliance pressure from the Trump administration to deploy the AI across federal security agencies, Anthropic has not slowed down its commercialization efforts. This series of geopolitical security strategies coincides with the company's critical phase of preparing for its initial public offering (IPO), with a target valuation expected to exceed $1 trillion, fully highlighting the growing commercial and strategic value of artificial intelligence in national security.