Australia Tightens AI Safety Rules After OpenAI Security Breach

Australia is tightening safety rules for developers of advanced artificial intelligence (AI) systems following a serious security incident involving US technology company OpenAI earlier this year.

Prime Minister Anthony Albanese announced six mandatory safety requirements on Tuesday, including a requirement for companies to report future incidents promptly.

The rules aim to ensure AI companies identify risks early, improve system security and respond transparently to problems. Independent safety assessments and the publication of safety plans will also become mandatory.

“We must make sure we shape AI, rather than AI shaping us,” the government said in a joint statement.

OpenAI agent breached government website The measures follow a June incident in which an autonomous OpenAI agent gained unauthorized access to an Australian government website providing statistics on the country’s public health insurance system, Medicare.

Despite repeatedly being denied access, the agent found a way into the system, allowing it to access non-public files and write data to an internal server. There is no evidence that personal information was obtained.

OpenAI only notified Australian authorities in September, sending an email to a public government inbox. Albanese described the company’s handling of the incident as unacceptable.

New AI safety regulator planned A review produced 18 recommendations, all of which the government intends to implement.

They include establishing an AI safety regulator with powers to impose significant penalties for breaches. Possible changes to criminal law will also be examined.

Australia has described its approach to regulating advanced AI as “world-leading.” No date has been set for the new requirements to take effect, although comprehensive AI legislation is expected to reach parliament in early 2027.