L051 · Staying safe · 2 min
Jailbreaks: talking AI past its rules
A jailbreak is a prompt written to trick a chatbot into ignoring the safety rules its makers gave it.
01 · The idea
Chatbots come with guardrails. They are trained and instructed to refuse things like weapon instructions, malware or help with harming someone.
A jailbreak tries to slip past those rules with a trick: a role-play setup, a fake emergency, a story inside a story, or a long confusing message. Some tricks work for a while, then the makers notice and patch them.
Companies hire people to attack their own models on purpose before release. This is called red-teaming, and it is how many weak spots get found and fixed.
For everyday users, the lesson is simple. Guardrails are useful but imperfect, so a chatbot saying something does not make it safe or allowed. And sharing jailbreak tricks can break a service's terms and get your account closed.
02 · Think of it like this
It is like a shop assistant trained to say no to certain requests. A smooth talker might sometimes get a yes, but that does not change the shop's rules, and the manager will retrain the staff.
03 · Do it now
04 · Quick check
Words in this lesson
See also
L185Under the hood · 3 min
When AI products fail in public
Public AI failures usually follow a few patterns, like invented facts, promises nobody approved and bots tricked into embarrassing replies, and each has a known fix.
L169Under the hood · 2 min
Guardrails on the way in
Input guardrails check what users send before the model sees it, catching private data, abuse and tricks early.
L017Staying safe · 2 min
AI and your kids
Children already use AI, so the goal is not to ban it but to teach them to use it wisely and safely.
L026Under the hood · 2 min
AI agents: AI that takes action
An agent is AI that does not just answer but takes steps: it plans, uses tools, checks the result, and keeps going.