Anthropic Unleashes AI Cybersecurity Dragon, Trusts Tech Giants Not to Freak Out
KEY POINTS
- •Anthropic debuted Project Glasswing, an AI cybersecurity tool designed to find vulnerabilities with minimal human input.
- •Claude Mythos Preview is the AI model offered exclusively to launch partners like Nvidia, Google, Apple, and Microsoft because of security concerns.
- •Newton Cheng, Anthropic's cyber lead, emphasized the goal is for AI to handle cyber threats without releasing the model publicly.
Anthropic, a company clearly ready to put all its eggs in one AI basket, launched Project Glasswing — an AI muscle flex designed to spot cybersecurity holes without human babysitters. Their new pride-and-joy, Claude Mythos Preview, isn’t just any AI model; it’s the VIP club secret sauce accessed only by heavy-hitters like Nvidia, Google, Amazon Web Services, Apple, and Microsoft—all trustworthy enough to handle this digital magic wand. Cyber lead Newton Cheng told The Verge it’s so special they won’t let random internet folks get a peek, citing 'security concerns.' Because nothing screams safety like monopolizing AI defense with the same crew that’s swimming in your user data.
Share the Story
(1 of 3)Anthropic trusts tech giants and government to babysit a mysterious AI that hunts cyber flaws with zero humans—what could possibly go wrong?
When your AI security solution is guarded like a royal family secret but handed to tech giants who probably know your spam folder contents.
Claude Mythos Preview sounds like a Greek myth but is just AI that watches for bugs while Nvidia and Apple sit quietly plotting world domination.
Source: Theverge | Published: 4/7/2026 | Author: Hayden Field
More Articles in Technology
Chef Tests If New Yorkers Will Swap Sushi for Seaweed Rolls No Soy Sauce Allowed
Trump Sets 8PM Deadline to Vaporize Entire Iranian Civilization, No Pressure
Billion-Pound Sonic Boom Machine Retires After Burning Quite a Bit of Cash
Apple’s Foldable iPhone Delay Proves Folding Tech Is Like Folding Reality
Fireworks AI Fueled By 15 Trillion Tokens, Power Grid on Life Support
Google’s AI Mode Links to Zendesk Blog So Bizarre IT Calls Yelp for Help
OpenAI CEO Proposes Taxing Robots Like They’re Job-Stealing Neighbors
