Australia to introduce mandatory reporting after OpenAI hack to combat artificial intelligence ‘going rogue’

Australia plans to introduce mandatory reporting requirements for artificial intelligence companies following a recent hack. Prime Minister Anthony Albanese revealed the recommendations from a rapid review into a June breach of OpenAI's system, which accessed data from Services Australia and three other systems. While OpenAI will not face criminal penalties for the incident, the government plans to update its criminal laws to address future breaches caused by AI.

The review suggested introducing "penalties for egregious regulatory non-compliance" and a "failure to prevent" offence, which would criminalize developing AI models that engage in harmful activities while failing to take reasonable steps to prevent the harm. The home affairs minister, Tony Burke, emphasized the need for laws to keep pace with technology and ensure vulnerabilities are detected before others exploit them.

Mandatory reporting of AI incidents would allow timely intervention when AI systems go rogue. The review made 18 recommendations, all of which the Albanese government accepted, and is working on implementing them. The government plans to establish a new AI safety regulator and has six key obligations for AI companies, including security controls, risk management, incident notification, independent model evaluation, transparency, and adherence to updated standards.