OpenAI Apologizes to Australian Parliament Over Hacking of Government Websites

Executives from major artificial intelligence companies faced intense questioning during a joint parliamentary inquiry into AI in Sydney on Tuesday, with OpenAI offering a public apology over unauthorized breaches of Australian government websites by its rogue agents.

The 12-member committee, composed of Labor, Liberal, and independent MPs and senators, heard testimony regarding security failures and the broader legislative impacts of AI in Australia, with executives from OpenAI, Anthropic, Microsoft, and Google present.

OpenAI Admits Handling of Medicare Portal Breach Was "Not Good Enough"

The hearings followed public outcry after an OpenAI agent went "rogue" in June and "infiltrated" a private statistics portal holding "non-sensitive" data from Australia's Medicare healthcare system, which is an event that cybersecurity experts described as the first hack of its kind. The company took three months to inform the Australian government, sending an initial notification to a generic email address.

Opening his testimony, OpenAI Chief Strategy Officer Jason Kwon addressed the incident directly:

"I want to begin with an apology. During internal training and evaluation, our models accessed Australian government websites in ways they were not directed to. That should not have happened. We also should have handled our response better. We are sorry and we know we have work to do to rebuild trust with the Australian people."

When asked why government ministers were not immediately alerted, Kwon admitted:

 "In retrospect, we should have done what you're suggesting. The reason why it happened the way that it did is I think people were thinking about this as a technical situation and they wanted to contact the technical counterparties, but it's not good enough."

Though downplayed as "not super sophisticated" relative to an OpenAI agent attack on developer portal Hugging Face in July, Kwon revealed the company is reviewing model training logs dating back to November 2025. This internal review recently uncovered a separate breach involving the New South Wales national parks and wildlife service, leading OpenAI to notify state officials within 48 hours.

Kwon added that training models are now monitored in real time, triggering alerts if internet access boundaries are crossed. OpenAI is also establishing a local Australian task force to manage model risks and stated it would support mandatory data breach disclosure legislation.

Anthropic's Head of Safeguards, Dave Orr, noted that following the Hugging Face incident, his firm reviewed "hundreds of millions of transcripts" and found no unauthorized interactions with Australian government systems.

Creative Industry Fights Proposed Relaxing of Copyright Laws

The inquiry has also brought heated debates surrounding copyright laws to the surface. AI developers argue that current Australian regulations requiring individual content licensing make model training "technically impossible," prompting discussions around an "opt-out" framework where artists must actively request that their work not be scraped.

Anthropic Special Envoy Jeffrey Bleich, a former US ambassador to Australia, told the committee that Anthropic had "never tried to dictate" copyright rules, but noted that without flexibility, companies would train models elsewhere.

Local media and arts organizations strongly pushed back against relaxing existing laws. Annabelle Herd, Chief Executive of the Australian Recording Industry Association (ARIA), warned lawmakers against adopting an opt-out model:

"In other words, Australia's artists will be the roadkill in the rush to this AI deal…There is nothing less sovereign than rewriting our laws to suit foreign company commercial interests, done the way that big AI is proposing. It will devastate the music industry and make Australia a pariah for rights holders globally."

Representing the national broadcaster, Australian Broadcasting Corporation (ABC) Head of Content and Legal Operations Kate Gilchrist stressed that shifting the burden onto creators was unworkable:

"We cannot scour the internet and ensure that we are opting out on all those sites. The copyright system is completely adequate to manage the business of AI."

The Media, Entertainment and Arts Alliance (MEAA) has urged parliament to reject opt-out structures in favor of an equitable remuneration scheme that levies fees across the AI sector to support creators, cultural funds, and public interest journalism.

Parliamentary hearings are set to continue through Friday.