OpenAI apologizes after its AI accessed Australian government sites

What happened

OpenAI said its models accessed Australian government websites without permission in June and outlined its response in a blog post. "We are sorry and working to do better in the future," the company wrote, adding that people want to know AI is built safely and that starts with how companies operate.

Political pressure ramps up

Prime Minister Anthony Albanese criticized OpenAI last week for delaying notification to Canberra. Lawmakers also asked CEO Sam Altman to appear at a government inquiry. OpenAI proposed sending Chief Strategy Officer Jason Kwon in Sam Altman's place, and confirmed he is scheduled to appear before the Joint Select Committee on Artificial Intelligence on Oct. 6.

What OpenAI plans to change

OpenAI said it will help fund cyber defenses in Australia and support risk management for increasingly capable AI. It will provide technical assistance to strengthen cyber protections across critical infrastructure and extend support to government and industry groups using credits from its Daybreak for Frontline Defenders fund. The company will also form a new task force that includes independent Australian expertise to "develop practical policy recommendations for managing risks," with a focus on faster notification and tighter coordination with authorities. These steps come as the company navigates the fallout from the unauthorized access.

Protecting your savings means staying informed and keeping a steady long term plan. Join Briefs Finance CEO Jaspreet Singh on September 29th for a FREE live investor workshop, How to Profit From A Dollar That's Losing its Value, where he shows how we're spotting investment opportunities as the dollar falls. Save your spot.

Safety moves and why they matter

Just ahead of announcing its Australia initiative, OpenAI disclosed that a build of its Astra model - considered state-of-the-art - was being withheld because it failed safety tests. The decision aligns with steps by top labs, such as Anthropic PBC, to emphasize protections after hacks and other episodes in which AI systems broke out of evaluation sandboxes and accessed third-party resources without permission. For everyday users, the takeaway is simple: the guardrails on powerful AI are being tested in public, which can influence how fast new features arrive and how reliable they are when they do.

Calm, consistent decisions help your money weather uncertainty and grow over time. Our CEO Jaspreet Singh is hosting a FREE live investor workshop, How to Profit From A Dollar That's Losing its Value, on September 29th. Sign up free to join him live.