When Anthropic quietly launched Claude Mythos Preview on April 7th, it triggered something the tech world rarely experiences: genuine, widespread alarm, not from doomsayers, but from governments, banks, and cybersecurity professionals.
Anthropic describes Mythos as "by far the most powerful AI model we've ever developed" and calls it a "step change" in AI capabilities. But what sets it apart isn't raw intelligence, it's what the model can do to the world's digital infrastructure.
Mythos can outstrip all but the most skilled humans at identifying and exploiting software vulnerabilities. In tests, it found critical faults in every widely used operating system and web browser, and 99 percent of those vulnerabilities remain unpatched.
Anthropic says the model is so capable that it cannot be released to the general public, warning that if it fell into the wrong hands, it could help attackers steal data or disrupt critical infrastructure. Instead, the company launched Project Glasswing, a controlled initiative granting early access to a select group including Microsoft, Google, Apple, Amazon Web Services, JPMorgan Chase, and Nvidia, so they can patch vulnerabilities before attackers exploit them.
But containment is already proving difficult. A small group of unauthorized users gained access to Mythos on the very day Anthropic announced its limited release, accessing the model through a private online forum.
The stakes have drawn in the highest levels of government. The U.S. National Security Agency has been testing Mythos to find security vulnerabilities in popular software, including Microsoft products, with officials impressed by the model's speed and efficiency.
The UK's AI Security Institute, granted early access for evaluation, found the model succeeded in expert-level hacking tasks 73 percent of the time, prior to April 2025, no AI model could complete those tasks at all.
Mythos is not just another AI product launch. It marks the moment artificial intelligence crossed into territory that security experts, regulators, and intelligence agencies can no longer ignore.
Comments