Mythos: Corporate weaponry

In this episode of De Dag, I explain what Mythos actually is and why the panic is partly justified. The worst-case scenario is real: an actor with access to Mythos could take down power grids, hospitals, logistics, aviation, payment systems — essentially anything that runs on electricity. You'd be thrown back to the Stone Age.

But the deeper issue isn't the model itself — it's who controls it. Anthropic has essentially built a cyber weapon and then decided on its own who gets access: about twelve major US companies like Apple, Cisco, and Palo Alto Networks. That's a corporate version of a ministry of cyber defense, without any democratic mandate. A company's purpose is profit, not public safety.

This is the paradox at the heart of Anthropic: they're simultaneously the most safety-oriented AI company and the biggest accelerator, constantly pushing out new models. They're hitting the brake and the gas at the same time. And the fundamental challenge of the 21st century is how we bring democratic control back to technology development. Perhaps what we need is a NASA for AI — technology that serves a public purpose, with democratic accountability.