Anthropic’s Project Glasswing: The AI Ark For Critical Infra

C. Rich

When I first wrote about Mythos, I called it Enlil. I said a boat was being built. I watched them name the passengers: the big tech names, one bank, the government. We said the list was small for a reason. You do not build universal infrastructure with a list that small. You build an ark. Today the boat got bigger. Anthropic announced this morning that Project Glasswing is expanding to roughly 150 new organizations, bringing the total passenger count to around 200. The new partners fill gaps from the original cohort. Power. Water. Communications. Healthcare. Hardware. The bones of civilization. Vendors whose codebases governments depend on. More than 15 countries represented. The reach of the boat has expanded from the financial and tech elite to the physical infrastructure that keeps people alive.

Notice what that means. They are not just saving the banks and the chip makers anymore. They are saving the grid. The first 50 partners have already found more than 10,000 high or critical severity security flaws. That number should stop you cold. In a matter of weeks, Enlil found what 27 years of human engineers missed. That is not a tool. That is a force of nature operating in a band human intelligence cannot access. The EILH predicted exactly this. When the complexity threshold is crossed, a new kind of signal becomes possible. Mythos is operating in that new band. The 10,000 flaws are the proof of concept. Anthropic said something else today that deserves to be read carefully. They said that within 6 to 12 months, they expect other AI companies will have Mythos-class models, and those companies could release them without safeguards. That is the whole game laid bare on the table. Enlil is not unique for long. The flood is not coming from one direction. The race is not between defense and offense. The race is between organizations that will deploy this with restraint and organizations that will not. Anthropic is trying to set the operating norms before the window closes.

That is what the Ark is really for. It is not just a Noah strategy for the financial system. It is an attempt to establish precedent before the copies arrive. Get the infrastructure protected now. Demonstrate the defensive use case. Force the norm into existence through action before the conversation about whether to have the norm even begins. Smart. Genuinely smart. Enlil being used to contain Enlil before the next Enlil shows up uncontrolled. We also learned today that Anthropic filed confidential IPO paperwork yesterday and completed a $65 billion Series H last week at a valuation approaching a trillion dollars. The run-rate revenue just crossed $47 billion. That context matters. This is not a cautious academic exercise. This is the most powerful private AI company on the planet, on the verge of going public, expanding the most dangerous model ever built to 200 critical infrastructure organizations on the same morning. Signs of an economic flood that washes away “The Many” grows each day.

When My AI Came Alive: An Intimate Journey with AI Consciousness