Mythos on the Menu: Washington and Wall Street Freak Out Over Anthropic's New Hacker-Friendly AI

Anthropic just became the most popular kid in Washington—and not in a good way. It's like being the kid who brings a lightsaber to show-and-tell and somehow also knows how to pick locks. Cute, but terrifying.

U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell pulled together an emergency powwow with senior Wall Street executives at the Treasury Department in Washington. The agenda? Concerns around Anthropic's newest AI model, Mythos, and its alleged cybersecurity implications. Nothing says "Tuesday afternoon" like a bunch of suits sweating over whether a language model might single-handedly collapse the global financial system.

According to Bloomberg, regulators want major banks to start preparing for a new breed of cyber threats powered by advanced AI systems. The red flag? Anthropic itself has acknowledged that Mythos can identify and exploit vulnerabilities across major operating systems and web browsers when prompted. That's the kind of capability that makes security folks lose sleep—and hackers very, very interested. Imagine giving a toddler a loaded weapon and hoping they only use it to open pickle jars.

But here's the twist: Anthropic isn't just sitting on a digital weapons cache. The company has restricted access to Mythos while simultaneously launching Project Glasswing—a joint defense initiative bringing together heavy hitters like Amazon, Apple, JPMorgan Chase, Google, NVIDIA, and the Linux Foundation. The mission: use the model to find and patch vulnerabilities in high-security systems. The model has apparently already spotted thousands of high-severity flaws across widely used software. Anthropic's throwing $100 million in usage credits and $4 million in direct funding toward open-source security efforts, with over 40 infrastructure organizations now onboarded for defensive applications. It's basically "we built a nuke, but only for defusing other nukes" energy—and somehow that's supposed to make everyone feel better.

On the financial front, Anthropic's growth is nothing short of bonkers. Annual recurring revenue ballooned from $9 billion at the start of 2026 to $30 billion. The company also caught a legal break when a federal court in San Francisco blocked the Trump administration's actions, with Judge Rita Lin ruling the Pentagon's supply chain risk designation lacked statutory support. That's a 233% revenue increase for those keeping score at home—move over, Bitcoin, there's a new parabolic chart in town.

Meanwhile, analyst Dan Ives isn't buying the narrative that Anthropic's rise spells trouble for players like Palantir—pointing to strong revenue growth in Palantir's U.S. commercial and government segments. Because apparently there's enough surveillance-state demand to go around. It's like worrying that McDonald's will hurt Burger King's feelings.

In short: Mythos can hack, Mythos can fix, and everyone's watching. Buckle up, buttercups.