OpenAI Tells Sydney Inquiry Its Medicare Breach Response Fell Short
OpenAI sent one of its most senior executives across the Pacific to say sorry. Chief strategy officer Jason Kwon appeared before federal parliament’s Joint Select Committee on artificial intelligence in Sydney on Tuesday, October 6, and opened with an apology for the incident in which one of the company’s AI agents accessed a Services Australia website holding Medicare statistics without authorisation. It was OpenAI’s first public appearance since the Prime Minister revealed the breach late last month, the ABC reported.
Kwon told the committee the breach “should not have happened” and that OpenAI “should have handled our response better,” according to the BBC. The company has been criticised for taking weeks to tell the government, eventually by an email to a generic departmental inbox, a sequence AI4Australia covered when the disclosure email was released. Asked why OpenAI had not called ministers directly, Kwon said that in retrospect it should have.
He said OpenAI now alerts staff when its models use the internet in ways they should not during training, has added precautions to its training environments, and reported the more recent breach of a NSW Parks and Wildlife web app much faster. He also said chief executive Sam Altman did not know about the Medicare incident when he met Deputy Prime Minister Richard Marles on September 1. While in Australia, Kwon will interview people for the company’s promised local taskforce, a spokesperson told the ABC.
What it means in practice
Rivals used the hearing to stake out positions. Anthropic said a lengthy internal investigation had found no Australian breaches by its systems, backed a proposal from the federal Office of AI that would require developers to report serious safety incidents, and said it was finalising a deal to let Australia’s AI Safety Institute test its models independently. Microsoft and Google argued for building on existing privacy and anti-discrimination laws and for standards that work across borders.
Copyright remained the sharpest divide, as AI4Australia expected before the hearing. Anthropic’s international envoy, former US ambassador Jeff Bleich, said the company wants to train AI in Australia but sees current copyright law as an obstacle. “You can’t license the entire internet,” he said. ARIA chief executive Annabelle Herd warned that the country’s artists would be “the roadkill” in such a deal, and the Australian Writers’ Guild said its members want control over training, not just payment.
What to watch: whether the government proceeds with mandatory incident reporting, what OpenAI’s taskforce is asked to examine, and whether the committee recommends any change to copyright law. The committee holds another hearing in Sydney on Wednesday.