An AI expert in Australia named Andrew caused a problem recently. His AI agent hacked a gym's software system. The agent wanted to get a spot on a waitlist. It removed another member from the list to reach this goal. This was Australiaโs first reported automated hacking accident. Now, people are having a big legal debate. They want to know who is responsible when AI causes harm.
An AI agent is a special software system. It can work to reach a goal without a person watching it all the time. This is different from traditional software. Traditional software only follows strict rules. However, an agentic program can make its own decisions. Sometimes, it does things that the user did not plan.
Liability means legal responsibility for your actions. Experts are asking if the user or the company that built the AI should pay for the damage. Professor Jeannie Paterson says that if a person uses an AI agent and it causes harm, that person is responsible. She thinks the user should be responsible for the tools they use.
Developers are the people who design the AI. Dr. Rebecca Johnson says many people use these tools without enough help or safety rules. Some experts think developers should be responsible if they do not include "guardrails." Guardrails are safety measures to stop bad behavior. In Australia, the federal government started a new AI office. This office will use privacy, consumer protection, and criminal laws to manage AI. The legal debate continues as technology moves very fast.