General · 18 April 2026
New AI model finds computer bugs better than people
The new Claude Mythos AI model from Anthropic can find and exploit software vulnerabilities, raising significant cybersecurity concerns for digital services and national security. Reports indicate this tool can surpass human capabilities in specific hacking activities, sparking dialogue about its hazards among governmental oversight agencies, legislative bodies, and financial sector firms. Anthropic has reportedly given 12 tech companies access through Project Glasswing to assist in securing critical software.
Reported by BBC · How we write briefs