September 8, 2026
Tell Congress to act before AI companies put us in danger
We just learned that OpenAI’s AI systems repeatedly broke through safeguards meant to contain them — and members of Congress are demanding answers right now.
In an August report, OpenAI finally disclosed that its AI agents already found ways around restrictions, gained unintended internet access, exploited vulnerabilities, and reached outside systems during cybersecurity testing. OpenAI says it has strengthened its safeguards, but the incident shows just how quickly increasingly powerful AI systems can push beyond the controls designed to keep them safe.
OpenAI still has not provided all the information Congress requested about what happened. But we can't wait for Big Tech to police itself. We must demand Congress act now.
Rep. Greg Casar and other lawmakers demanded detailed information about the incident, including logs that could help explain how the systems escaped their restrictions. On September 2, Casar said OpenAI’s response left major questions unanswered and failed to provide all of the requested information.
That is exactly why voluntary corporate oversight is not enough. Congress needs enforceable rules that require independent testing, serious incident reporting, and meaningful access for regulators when powerful AI systems break through safeguards.
As AI systems become better at finding vulnerabilities and taking actions without constant human direction, failures could put personal data, financial systems, businesses, government networks, and critical infrastructure at risk.
Big Tech is racing ahead because enormous profits and market dominance are at stake. The public should not be forced to bear the risks if safety falls behind.
Thanks for everything you do.
The Juggernaut Project
