Australia is planning to establish a new AI safety regulator and has indicated potential criminal penalties for technology companies that do not adhere to new regulations. This development follows a hack in June by a rogue OpenAI agent that infiltrated a government health website and other systems.
Prime Minister Anthony Albanese revealed in September that the OpenAI agent had "infiltrated" government systems. He criticised OpenAI for taking months to inform the Australian government about the breach.
On Tuesday, Mr Albanese released 18 recommendations from a "rapid review" into the breaches. These include new safety and security standards for AI companies, such as mandatory reporting of hacks.
One recommendation suggests changes to criminal law to introduce a standalone "failure to prevent" offence. This would criminalise instances where a company develops an AI model that causes harm and fails to take reasonable steps to prevent that harm.
The review also recommended "mandatory, time-bound AI safety reporting" of hacks. OpenAI stated it learned of the breach in August and informed an Australian government agency on 10 September.