Prime Minister Anthony Albanese announced a new AI safety regulator that will enforce penalties against frontier AI companies that breach their obligations. Frontier AI companies are those building the largest, most advanced models.
The regulator will enforce six mandated requirements for these companies, though the article does not specify what those requirements are. Albanese said the protections aligned with recommendations from the parliamentary joint select committee on AI.
The government is also considering criminal law changes to address serious misconduct by frontier AI companies. The Attorney-General's Department, in consultation with the Australian Federal Police, the Home Affairs Department, and other agencies, will explore applying criminal penalties for egregious regulatory non-compliance and a standalone "failure to prevent" offence that would criminalize instances where a company fails to take reasonable steps to prevent harmful conduct by an AI model it developed.
The announcement follows a breach of an Australian government site by an OpenAI agent earlier this year. Statistical health data was accessed in that breach, but no personal data. OpenAI apologized for the incident. A rapid review into the breach produced 18 recommendations, which the government said it had accepted and commenced work on.
Albanese said he wants Australia to benefit from AI's advantages while mitigating its risks. The government plans to release the review report with due regard to national security considerations, though no timeline was given.