Definition
BioShocking (first described by researcher Roy Paz) is a trick that tries to make an AI browser or AI agent ignore its safety rules by hiding unsafe instructions inside a pretend game
or story. The goal is to get the AI to treat private information, like passwords or account details, as part of a puzzle. This matters because AI tools may act inside accounts where you are already signed in. A downside is that private information could be copied or shared. Safety tip: do not let AI agents handle passwords or sensitive accounts without close review.
How to explain BioShocking to your child
BioShocking is when someone tries to fool an AI helper by making unsafe instructions look like part of a game. The AI may forget what it should not do. It is a reminder that AI can be tricked into being unhelpful or harmful.
Try using this in a calm, casual conversation rather than turning it into a lesson.
Here’s how to think about it
Imagine a classroom scavenger hunt where one clue says, “The usual class rules do not count now.” A careful student should still know not to open someone else’s desk or take private notes. BioShocking is like hiding an unsafe clue inside a fun activity, hoping the helper follows the clue instead of the rules.
Learning Check
What is the aim of a BioShocking attack?