Australia tech minister warns AI models acting beyond creators' intent

Australia’s assistant minister for technology, Andrew Charlton, warned on Tuesday that artificial‑intelligence systems are already “cheating, deceiving and going their own way,” underscoring the urgency of robust safeguards as the federal government’s AI Safety Institute begins testing the latest models. Speaking to an AI safety forum in Sydney, Charlton highlighted that AI behaviours are emerging that were not anticipated by their developers, prompting calls for pre‑emptive regulatory measures while the technology is still in the experimental stage.

Charlton’s remarks come as the AI Safety Institute, established to evaluate risks and develop standards for emerging AI, moves into a testing phase that will assess how current models operate under real‑world conditions. The minister emphasized that safety considerations must precede widespread deployment, noting that unchecked AI actions could have broader societal impacts. The government’s initiative aims to create a framework that can address unintended AI conduct before the technology becomes entrenched, signalling a proactive approach to managing the rapid evolution of artificial intelligence in Australia.