IT Home On November 10, according to Reuters, Australian Prime Minister Anthony Albanese said on Tuesday local time that Australia will require artificial intelligence companies to report security violations and consider amending criminal law to address “serious misconduct” in the AI sector.
This decision stems from a security incident that occurred in September this year: an AI agent from OpenAI accessed the Australian government website without authorization.
According to previous disclosures by the Australian government, the AI agent entered the medical statistics portal of the Medicare health system without permission while studying public healthcare expenditures. Medicare is Australia’s universal health insurance program. This incident subsequently prompted the Australian government to quickly initiate a special review.
Albanis said that the government has accepted all the suggestions made during the review.
He said in a media interview: “These suggestions include requiring AI companies to mandatoryly report when their behavior deviates from expectations or when there are cybersecurity incidents, as well as considering reforming criminal laws for serious misconduct.”
He also said, “Laws need to keep up with this rapidly changing world, and in some cases, it is necessary to ensure that laws can keep pace with technological developments.”
IT Home noticed that OpenAI previously stated that the company only learned about this incident in August, and emphasized that the unauthorized access was not intentional. Internal investigations by the company found no evidence that the AI agent accessed patient medical records.
This incident further intensified the tensions between Australian and US technology companies.
Previously, the Australian government refused to grant copyright law exemptions to OpenAI, the developer of ChatGPT, and its competitor Anthropic. This prevented the two companies from using copyrighted content without authorization when training AI models. Australia required these companies to negotiate content licensing agreements with local copyright holders.
However, last month, both companies stated at the Australian Parliament hearing that they are willing to comply with legal requirements that AI companies report data leakage incidents caused by their agents. The two companies admitted that whether to notify relevant authorities about such incidents is still up to the companies themselves.
Albanes said that the center-left Labour Party government under his leadership will establish an AI regulator responsible for overseeing safety measures throughout the entire process of AI system development and deployment, identifying major risks, and conducting independent evaluations of AI models.
He said, “AI safety regulatory agencies will have the authority to impose heavy fines on AI companies that engage in violations.”
Albanis said that mandatory AI safety and protection requirements will be included in the government's proposed AI legislation. Australia plans to submit the relevant bill to parliament for consideration next year.