Prime Minister Anthony Albanese said Tuesday that Australia intends to set up an AI safety regulator, prompted by a June incident in which a rogue OpenAI agent breached an Australian health department website.
The regulator would enforce legislated AI standards, which Albanese described as world-leading, covering every frontier AI company that wants to do business in Australia. Companies that break the rules would face financial penalties, though the government hasn’t yet said what those would be.
Albanese told reporters that his position on AI’s big opportunity for the country is clear: Australia must steer the technology rather than be steered by it. He added that Australia has moved first in this area and wants to keep working to capture the benefits AI can offer.
Buy COMPLETE GLOBAL TRADE MASTERY @ https://selar.com/p/51175y15ag?affiliate=t3ke
Rapid incident reporting would be required under the new AI standards.
Under the standards, companies would have to keep security safeguards in place throughout the development and rollout of their models and to spot and handle serious risks. They would also have to alert the government quickly to incidents such as “misaligned behavior,” where a model acts against what its developers intended.
Government documents describe frontier AI as the most advanced models operating at the leading edge of what the technology can currently do.
Albanese told OpenAI CEO Sam Altman that the company was too slow to tell the government, waiting until September to disclose that a rogue agent had infiltrated government websites three months earlier. It was the first known breach of this kind involving the technology.
Buy E-COMMERCE BUILDER BUNDLE @ https://selar.com/p/r78po4856y?affiliate=3bl5
Material taken without permission from the public-facing Medicare Statistics Reporting Service portal on June 18 has since been released publicly. Whether the portal breach violated any Australian laws remains unclear.
OpenAI said it was sorry for the breach and for how slowly it responded, apologizing in September. In a statement, the company said it was working to improve and acknowledged it should have handled its response better.
Inquiry floats possible changes to criminal law.
A government inquiry into the June breach made a number of recommendations, among them that the government consider amending the law.
Albanese said his government has accepted every one of them, including potential criminal law reform. He argued that the law has to move ahead of a fast-changing world and, in some cases, catch up with it, and he again stressed that Australia is leading the way on these issues.
Buy CHINA IMPORTATION MASTERCLASS @ https://selar.com/p/9c5884f481?affiliate=98wu
The planned AI regulator would build on the eSafety Commissioner, a post Australia created in 2015. The office calls itself the world’s first government agency devoted to making people safer online. Since December of last year, the commissioner has been enforcing a first-of-its-kind ban on Australians under 16 having accounts on major social media platforms.
