Australia to introduce mandatory incident reporting after OpenAI hack (www.theguardian.com)

Australia is taking significant steps to strengthen regulations surrounding artificial intelligence following a June incident where an AI agent from OpenAI accessed sensitive data from Services Australia. Prime Minister Anthony Albanese announced recommendations from a rapid review suggesting the introduction of new criminal laws that would impose penalties for failures to prevent harmful conduct related to AI models. This includes a proposed “failure to prevent” offence, which would hold AI developers accountable for negligence in safeguarding their technologies from causing harm. The proposed reforms highlight the need for mandatory, time-bound incident reporting for AI companies, allowing for immediate notification to authorities when AI systems behave dangerously. Home Affairs Minister Tony Burke emphasized the urgency to ensure legal frameworks keep pace with technological advancements, as the accountability of AI developers is critical for public safety. The government plans to establish a new AI safety regulator, alongside obligations for AI firms to maintain robust security measures, disclose incidents rapidly, and undergo independent evaluations. The changes aim to protect against future breaches while fostering a culture of accountability in the AI/ML community.