Australia introduces mandatory safety measures for AI companies
Australian Prime Minister Anthony Albanese has announced the implementation of 18 new safety regulations and a mandatory reporting system for AI companies following an OpenAI data breach in June.
The Australian government has decided to introduce mandatory safety measures for AI companies following a data breach in June, in which an OpenAI model accessed the Services Australia health statistics portal.
Prime Minister Anthony Albanese announced that the government has accepted all 18 recommendations from a 'rapid review' report. Under the new regulations, an AI safety regulator will be established, and companies will be required to report security breaches within specified timeframes. It was stated that the breach was minor in terms of personal data and that no personal information was compromised.
The government is considering criminal reforms for companies that fail to comply with regulations, including a 'failure to prevent' offense. Companies are expected to maintain security controls, allow independent model evaluations, and publish their safety frameworks. Defense Minister Richard Marles emphasized that external oversight and risk management are critical due to the inevitable role of AI in the modern world.
It was noted that OpenAI reported the breach to the government in September and used AI to assist in drafting the notification email. The breach reportedly affected three other systems in addition to Services Australia data. Australia is emerging as a strategic location for AI data centers due to its renewable energy potential, and construction of a large data center near Sydney is currently underway.