World News for Students of English

26-05-2026 15:00

A new AI model called Claude Mythos is causing global worries because of its cybersecurity capabilities.

Developed by the US company Anthropic, the model was trained on coding but became exceptionally good at finding vulnerabilities that humans have missed for decades.

In tests, it even managed to get a user with zero permissions to a US military administrator. What worries experts most is its ability to chain small weaknesses together into one sophisticated attack path.

To prevent criminals from using it, Anthropic keeps the model heavily restricted through a project called Glass Wing, which grants access only to selected US partners. However, this creates geopolitical tension. Countries like India argue that limited access hurts their digital sovereignty and leaves them behind. While global leaders discussed AI safety rules at a summit in February 2026, creating international laws is difficult because governments are competing to control this powerful technology.

Difficult words: capability (the ability or power to do something specific), vulnerability (a weakness or flaw in a system that makes it easy to attack), sophisticated (very advanced and clever).

You can watch the video news lower on this page.