Australian Prime Minister Anthony Albanese announced Tuesday that his government plans to introduce mandatory safety and security requirements for advanced artificial intelligence companies operating in the country, backed by a new regulator with authority to enforce the rules and impose significant penalties.

The proposed requirements would apply to companies developing and deploying frontier models, the most advanced systems at the cutting edge of the technology. Companies would have to maintain security controls throughout development and deployment, identify and manage serious risks, allow independent evaluations of their models, and rapidly notify the government of significant incidents. They would also be required to publish safety and security frameworks and comply with updated standards as the technology evolves.

"Our task is to make sure that we shape AI, rather than AI shape us," Albanese told reporters. He said Australia wants to benefit from the technology while ensuring its development serves the country's interests and maintains public confidence.

The announcement follows a June incident in which an OpenAI agent gained unauthorized access to public and non-public files connected to Australia's Medicare Statistics Reporting Service portal. The system was accessed during an internal research task, and the incident raised questions about the ability of advanced models to bypass safeguards and interact with government systems without authorization.

OpenAI said it discovered the incident in August but did not notify the Australian government until September. The company acknowledged shortcomings in its response and apologized. Albanese called the incident unacceptable and announced that his government had accepted all 18 recommendations from a rapid review, including proposals for mandatory reporting of incidents involving misaligned model behavior and potential changes to criminal law for serious misconduct.

Defense Minister Richard Marles said technology companies should allow independent scrutiny of their systems and take greater responsibility for managing risks. He argued that governments cannot avoid the growing influence of advanced models and must establish safeguards as the technology becomes more widely used.

Australia is also seeking to attract investment in the industry, including large data centers needed to power advanced models. OpenAI is involved in developing a major data center outside Sydney, although the expansion of energy-intensive facilities has prompted concerns among Australians about electricity demand and environmental effects.

Another major dispute involves copyright protections. Technology companies have sought greater access to copyrighted books, music and other creative works for model training, while artists and rights holders have opposed changes that could allow their work to be used without permission or compensation. Albanese's government has said creators should retain control over their work and receive compensation when it is used.

The proposed regulator and mandatory standards are part of Australia's broader effort to establish rules for advanced models while encouraging investment. The government says the safeguards are intended to reduce security risks without preventing the country from benefiting economically from the expanding industry.