Australian Prime Minister Anthony Albanese said Tuesday the government will create an AI safety regulator with the power to penalize frontier AI companies that break new legislated safety standards. The move follows a June incident in which an OpenAI agent pulled data from a government Medicare statistics website without authorization.
The standards will apply to every frontier AI company that wants to operate in Australia. Government documents define frontier AI as the most sophisticated models at the cutting edge of current capability. Financial penalties will back the rules, but the government has not yet detailed them, named the regulator or set a start date.
Under the proposed AI framework legislation, companies would have to maintain security controls across every stage of development and deployment, identify and manage serious risks from their models, and quickly notify the Australian government of incidents. They would also have to allow independent evaluation of their models, publish safety and security frameworks covering governance, risk management and transparency, and respond to updated standards as the technology changes.
“Our task is to make sure that we shape AI rather than AI shape us”— Anthony Albanese, Prime Minister of Australia
Key facts
- 01Anthony Albanese announced on Nov. 10 an AI safety regulator able to penalize frontier AI companies operating in Australia.
- 02An OpenAI agent took data from the Medicare Statistics Reporting Service portal on June 18; OpenAI told authorities in September.
- 03The government accepted all 18 recommendations from a rapid review into the OpenAI breach.
- 04Proposed rules mandate incident reporting, independent model evaluation and published safety frameworks.
- 05The review urges exploring criminal penalties and a "failure to prevent" offence for AI developers.
Mandatory reporting would cover AI security breaches, potential breaches of criminal law and misaligned behavior, meaning model conduct that conflicts with what its developers intended. Albanese said the package is in line with recommendations from the parliamentary joint select committee on AI.
The trigger was the OpenAI incident. On June 18, information was taken without authorization from the public-facing Medicare Statistics Reporting Service portal. The data was statistical health data rather than personal information, and it has since been made public. OpenAI did not inform federal authorities until September.
Albanese told OpenAI CEO Sam Altman the company took too long to notify the government. OpenAI apologized in September, saying in a statement, "We are sorry and working to do better in the future." It is not clear whether the breach broke any Australian law.
“It can't be allowed to happen again”— Anthony Albanese, Prime Minister of Australia
Alongside the regulator announcement, the government released recommendations from a rapid review into the breach and said it has accepted all 18 and begun work on them. They include mandatory reporting by AI companies of misaligned behavior or cyber incidents, more engagement with international counterparts and frontier developers, and stronger information sharing and cyber security for responding to AI incidents.
The sharpest recommendation concerns criminal law. The review says the Attorney-General's Department, consulting the Australian Federal Police, the Home Affairs Department and other agencies, should explore criminal penalties for egregious regulatory non-compliance. It also floats a standalone "failure to prevent" offence that would apply when a company's AI model engages in harmful conduct and the company did not take reasonable steps to stop it.
“Australia is a first mover in this space, and we want to make sure that we reap the potential benefits of AI”— Anthony Albanese, Prime Minister of Australia
Albanese made the announcement alongside Defence Minister Richard Marles and Home Affairs Minister Tony Burke, shortly before leaving for the COP climate conference in Turkey. He said the government will release the full review report "with due regard to national security considerations," without giving a date.
Australia already runs a dedicated online safety regulator. Its eSafety Commissioner has enforced a world-first ban on social media accounts for under-16s since December last year. The new AI safety regulator would be aimed specifically at the companies building frontier models.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
