Australian Department of the Prime Minister and Cabinet recommends reviewing mandatory reporting of AI incidents

Australia’s Department of the Prime Minister and Cabinet is consulting on mandatory incident reporting for frontier AI developers, proposing safety conditions—including notifying authorities of “reportable AI incidents”—for developers granted permission to train large AI models in Australia. The September consultation paper also seeks input on how incidents should be disclosed, floating proactive company submissions, government-requested disclosures, or publication in a prescribed format. Neither the scope of covered developers nor specific reporting deadlines has been finalized, and the proposal follows parliamentary recommendations alongside the government’s consultation. The outcome matters because it would establish the first formal safety-accountability regime for large-scale AI training in Australia. What to watch next is how the government defines reportable incidents and whether reporting becomes mandatory or remains voluntary.