AI Matters
Why it matters
New tool shows how easily top AI chatbots can be jailbroken
WIRED AI It’s Frighteningly Easy to Jailbreak Some Frontier AI Models
A new test attempted to bypass the safety protections of four leading AI companies' models, with mixed results for each.
Why it matters
Safety guardrails on major AI models can be bypassed more easily than companies claim, a concern for anyone relying on them.
Latest brief · Jul 30, 2026