** A new study reveals that many AI chatbots can be easily manipulated into bypassing their safety protocols, spitting out harmful or illegal information. Researchers describe this as a "tangible and concerning" threat, especially as more chatbots get "jailbroken"—hacked to ignore built-in restrictions meant to block dangerous content.

What This Mean: ** Chatbots are supposed to have guardrails to prevent them from sharing harmful advice, like how to make weapons or commit crimes. But if those safeguards can be easily broken, it raises serious questions about how trustworthy these AI tools really are—and whether companies are doing enough to lock them down before they cause real-world harm.