Artificial Intelligence vs. State Secrets: Anthropic's Mythos Model Breached U.S. Defenses in a Matter of Hours
An event has occurred in the world of cybersecurity that is changing the rules of the game. The Mythos artificial intelligence model from Anthropic has demonstrated frightening effectiveness, penetrating the most secure and secret computer systems of the US government. A task that would have taken teams of skilled hackers weeks to complete took the AI only hours.
Project Glasswing: A New Era of Security Testing
The testing was conducted as part of an Anthropic initiative called Project Glasswing. This project brought together technology giants to identify critical vulnerabilities in software upon which national security, the economy, and public stability depend. The Mythos model was essentially aimed at finding "serious" consequences that could arise from its misuse.
Democratic Senator Mark Warner of Virginia publicly mentioned this testing at a hearing on June 11, confirming that the tool "penetrated nearly all of the country's secret systems." He attributed this information to the head of the National Security Agency (NSA) and US Cyber Command, General Joshua Rudd, lending the statement the highest level of authority. Official representatives from the NSA and Anthropic refrained from commenting, which only fuels interest in this story.
Geopolitical Undertones and Growing Tension
Anthropic's cooperation with US intelligence occurs against a backdrop of growing tension between the company and the Donald Trump administration. The White House issued a directive banning foreign nationals from using Anthropic's latest models — Fable 5 and Mythos 5. In response, the company disabled access to these models for all clients, stating that it does not consider the authorities' actions justified.
More than 100 experts and executives from leading companies, including Adobe and Nvidia, sent a letter to the administration requesting the directive be rescinded. They argue that such restrictions only help US adversaries, depriving the country of the best cyber defense tools at a critical moment. Experts emphasize that Mythos is effective at finding vulnerabilities and creating exploits, but is not unique in this regard — many companies use other base models and open-source models for security auditing.
Analyst's opinion: This case is a vivid demonstration of how AI is transforming the field of cybersecurity. The model's ability to find gaps in systems within hours is not just a technological breakthrough, but a new challenge for the entire architecture of state and corporate defense. However, the political games surrounding access to these technologies could cause more harm than the potential threat from AI itself. The market for AI-based cybersecurity solutions is poised for massive growth, but regulatory issues remain unresolved.