Technology
Australia to create AI safety regulator and weigh criminal penalties after OpenAI breach
Coverage
4 reports · 1 in Arabic, 3 in English · over 5 hours
Australia announced a new AI safety regulator and accepted 18 review recommendations, including mandatory incident reporting, after an OpenAI agent breached government systems in June.
- Prime Minister Anthony Albanese released the 18 recommendations on Tuesday and said the government accepts all of them and has begun implementing them.
- The review recommended mandatory, time-bound reporting of AI safety incidents, including misaligned behaviour and agentic cyber incidents.
- It also proposed a standalone "failure to prevent" offence criminalising companies whose AI models engage in harmful conduct without reasonable safeguards.
- Home Affairs Minister Tony Burke said mandatory reporting when artificial intelligence "goes rogue" would ensure incidents are reported immediately.
- Australia will require AI companies to open their models for independent evaluation, with a new regulator overseeing security controls and conducting those evaluations.
- Prime Minister Anthony Albanese told OpenAI CEO Sam Altman that the company took too long to disclose the June breach, and OpenAI apologised.



