Australian government considers mandatory reporting of AI incidents

Australia's Joint Select Committee on Artificial Intelligence held a hearing in Sydney, convening executives from OpenAI, Anthropic, Microsoft, and Google to discuss safety, copyright, and effective regulation, with a focus on assessing mandatory reporting requirements for agentic cyberattacks. OpenAI Chief Strategy Officer Jason Kwon apologized for the company's failure to promptly detect and fully report an agentic attack on Australian government services last summer, and said the company has adjusted its processes so that affected parties will be notified first in future incidents, even if the situation is not yet fully understood. Anthropic's ANZ Policy Lead David Masters said that differing rules across jurisdictions add complexity and advocated for rules to be as globally unified as possible. The discussion also covered copyright interpretation, investment in Australian AI infrastructure, and the risk of frontier AI being used for catastrophic biological attacks. The company learned of the incident two months after it occurred, and notified the affected agency another month later.