Anthropic AI Found 27 Year Bug Exchanges Beg

Anthropic built an AI so good at finding security flaws that they're locking it away from the public—it discovered a bug in OpenBSD that hid for 27 years and thousands of zero-day vulnerabilities hackers would kill for. They're giving early access through Project Glasswing to about 40 elite organizations like Amazon, Apple, and Microsoft, but crypto exchanges like Coinbase and Binance are desperately lobbying to get in because deepfake attacks just jumped 783 percent and criminals already released AI tools that can beat their identity verification systems. We're watching a security aristocracy form in real-time where whoever gets the best defensive AI first survives, and everyone else becomes the next cautionary tale.