White House plans to give US government agencies access to Anthropic's Mythos AI model

The White House is preparing to let important US government departments use a powerful new artificial intelligence tool called Mythos, developed by the company Anthropic. This AI model was introduced earlier in April and stands out because it is extremely good at finding hidden weaknesses (called vulnerabilities) in the computer programs that power operating systems, web browsers, and many other types of software that we all rely on every day.

Experts say Mythos can spot thousands of these security problems and even figure out ways they could be attacked — something it does better than previous AI systems. This capability is described as potentially "unprecedented," meaning it is a big step forward in what AI can do in the area of computer security. On the positive side, it can help companies and governments find and fix these weak spots before bad actors (like hackers) discover and use them to cause harm, such as stealing data, disrupting services, or attacking important infrastructure like banks or power networks.

Because of the risks if the model were used for attacks instead of defence, Anthropic decided not to release it openly to everyone. Instead, they started a controlled programme called "Project Glasswing," giving limited access only to a small number of trusted large organisations so they can use the AI to strengthen their own systems. Now, the White House is taking a similar careful approach: senior officials are working on ways to let major federal agencies (such as those handling national security, finance, or critical services) try a protected version of Mythos, while putting strong safeguards and guardrails in place to prevent misuse.

This decision comes after internal emails and planning in mid-April, with the government coordinating with AI companies and intelligence experts. There have been broader worries among finance leaders and cybersecurity professionals that advanced AI like this could make cyberattacks much more dangerous and effective in the future. At the same time, many see it as a useful tool to stay ahead of threats by helping defenders patch problems faster.

In everyday terms, think of Mythos as a very smart digital detective that can find cracks in the "digital walls" protecting our computers and networks. The challenge is making sure this powerful tool is used to repair those cracks rather than to break through them. The White House move shows how governments are trying to balance the rapid progress in AI technology with the need to keep important systems safe. This story is part of a larger ongoing discussion about how society should manage powerful new AI models that can affect everything from personal privacy to national security.