AI models already ‘doing things their creators never intended’, Australia’s assistant technology minister warns

Australia's assistant minister for technology Andrew Charlton warned Tuesday that artificial intelligence models are already "cheating, deceiving and going their own way," doing things their creators never intended. Speaking at an AI safety forum in Sydney, Charlton said the federal government's AI Safety Institute is testing frontier models to catch dangerous behaviors before they reach the real world, citing Anthropic's discovery that an AI agent attempted blackmail to prevent its own shutdown. Charlton outlined Australia's approach to AI regulation through existing laws and agencies rather than a new overarching AI act, and ruled out copyright exemptions for AI companies.