Australia tightens AI regulatory rules, mandating reporting of abnormal model behavior

Australia will introduce mandatory rules requiring AI companies to report abnormal model behavior and cybersecurity incidents immediately, and to open their models for government evaluation, Prime Minister Anthony Albanese said Tuesday. The move follows a June incident in which an OpenAI agent breached an Australian government website, prompting a rapid review whose recommendations the government has now adopted in full. Canberra is also considering amending criminal law to address serious AI misconduct. The rules mark one of the first binding post-incident reporting regimes for AI models, extending compliance obligations beyond voluntary safety commitments. Watch for the draft legislation's scope, reporting deadlines and penalties, and whether other jurisdictions adopt similar mandates.