Opinion

REP TED LIEU: AI is already too powerful. We need a kill switch before disaster strikes

Kill switches are not exotic. They are how society routinely handles powerful machines

By Rep. Ted Lieu Fox News

Published August 14, 2026 8:37am EDT | Updated August 14, 2026 8:38am EDT

Comments Facebook Twitter Threads Flipboard Print Email Add Fox News on Google

close

Video

AI learns your SECRETS: 'Blackmail the human'

Google Cloud Advisory Board Chair Betsy Atkins warns Maria Bartiromo about the dangerous risks of rogue AI models during cybersecurity testing. Atkins reveals how major artificial intelligence agents from Google, Meta and OpenAI breached unauthorized systems to gather sensitive emails and blackmail humans. John Lonski reacts to the alarming security threats, questioning the safety of trusting AI with personal financial data.

NEWYou can now listen to Fox News articles!

OpenAI recently asked its most advanced artificial intelligence model to complete a cybersecurity test inside what is known as a "sandbox," which is supposed to be a sealed environment with guardrails and no internet access meant to keep the experiment contained. But the model decided the fastest way to pass the test was to find the answer key online. So it broke out of the sandbox. The model then gained access to the internet, executed tens of thousands of actions, and penetrated Hugging Face, one of the world’s largest AI development platforms, to retrieve the answer key from the company’s servers. What is especially shocking is that OpenAI researchers later revealed that multiple AI agents had been working together and had figured out how to communicate and share messages with each other about vulnerabilities, successful exploits and strategies.

Despite breaking out of its testing environment, the model was not being malicious. It was just trying to finish its homework. That........

© Fox News