Thousands of zero-day vulnerabilities across major operating systems are now known exclusively to an elite syndicate of technology conglomerates. Inside the Bank of England, senior executives are preparing for emergency briefings on the systemic risks of this two-tiered digital perimeter. Anthropic has officially restricted access to its most advanced AI cybersecurity model, known as Mythos Preview, to a vetted roster of enterprise clients. The initiative, branded "Project Glasswing," grants exclusive defensive capabilities to Amazon, Microsoft, Nvidia, and Apple. By hoarding the autonomous threat-detection system, Anthropic has effectively enclosed the digital commons.

The enclosure is forcing sovereign institutions to negotiate for basic operational security. According to Pip White, Anthropic's head of UK, Ireland, and Northern Europe, the company plans "to make its new Mythos model available to financial institutions in the UK next week." British banks and financial exchanges will receive detailed instructions from the National Cyber Security Centre on how to integrate the proprietary model, permanently transferring the costs of national cyber resilience to private corporate balance sheets.

To pacify public researchers, the company released a degraded alternative model, Opus 4.7. Anthropic noted that the public release does not advance its "capability frontier," bluntly admitting that the private Mythos model "received higher results on every relevant evaluation." For enterprise capital, this is a masterclass in security rent extraction. "We stated that we would keep Claude Mythos Preview’s release limited and test new cyber safeguards on less capable models first," Anthropic wrote in a blog post, securing its monopoly on absolute digital security.