With the bargepole close at hand, we dug into the news from Anthropic that it had launched Project Glasswing, to pre-empt an onslaught of AI-based cyberattacks. For the telco industry, with its mountain of collective technical debt, stretched operational support budgets, and chronic brain drain, Glasswing should serve as a major wakeup call. The pitch is that Anthropic’s new but still unreleased model, Claude Mythos Preview, has spent the past few weeks finding thousands of previously undisclosed zero-day vulnerabilities, including one introduced into OpenBSD over 27 years ago, several that are in the Linux kernel, and one in the ffmpeg video encoder that has apparently not been found despite being subject to 5 million previous automated scans. Project Glasswing hopes…