The incident dates back to June, when OpenAI's models accessed four Australian government websites in ways they were not authorized to during internal training and evaluation. Canberra was not notified until September, when the company sent an email to a public mailbox.
Prime Minister Anthony Albanese revealed the breach last week and rebuked the company for the delayed disclosure.
We Are Sorry and Working to Do Better
In a statement issued on Tuesday, OpenAI admitted its response had fallen short. "We also should have handled our response better. We are sorry and working to do better in the future," the company wrote, calling the breach "a new kind of cyber incident which represents an emerging global challenge."
The company said it will create an Australian task force, including independent Australian expertise, to "develop practical policy recommendations for managing risks" from increasingly capable AI agents.
The task force will focus on better notification and coordination with the government and is expected to complete its recommendations by year-end. "People want to know AI is being developed safely, and that starts with what companies like ours do ourselves," OpenAI wrote.
$1 Billion in Cyber Defense Credits
Alongside the task force, OpenAI pledged to support Australian governments and industry with credits from its $1 billion "Daybreak for Frontline Defenders" fund, plus technical assistance to strengthen cyber defenses across critical infrastructure.
OpenAI said it will also share technical findings with the affected agencies and set up an information-sharing mechanism so response teams can reach it directly.
Daybreak was launched earlier in 2026 as a gated service for verified defenders. In early September, OpenAI pledged $1 billion in subsidized access, training, and support aimed at frontline organizations protecting essential services, starting with U.S. utilities, local governments, and community banks.
Australia is now its first wartime-adjacent partner to receive the extension.
Chief Strategy Officer Jason Kwon Heads to Parliament
Australia had called on OpenAI chief executive Sam Altman to attend a government inquiry, and the company responded by offering its chief strategy officer instead. Jason Kwon will travel from the United States to Australia to answer questions at a hearing of the Joint Select Committee on Artificial Intelligence in Sydney on October 6.
OpenAI separately disclosed that its models also accessed publicly available information from U.S. government websites, including those of the Census Bureau and the Securities and Exchange Commission — a disclosure that is likely to keep regulators on both sides of the Pacific pressing the company for answers well beyond the October hearing.
By Hamza Faisal
