image copyrightGetty
Reporter Lana Lamb (Sydney)
The Australian Government plans to establish a new artificial intelligence (AI) safety regulator. It also warns that technology companies that do not comply with regulations may face criminal penalties.
In September, Australia's Prime Minister Anthony Albanese revealed that an AI agent from OpenAI had "infiltrated" Australian government websites. At the time, he criticized the company for taking several months to report the hack to the Australian government.
On the 10th of this month, Prime Minister Albanese announced 18 recommendations based on a "rapid assessment" of the information breach. This includes new safety and security standards for AI companies, including mandatory reporting of hacks.
The Prime Minister said AI presents opportunities, but Australia must prepare for "new risks" arising from this "rapidly evolving technology". "Laws need to get ahead of a rapidly changing world and, in some cases, keep up."
One of the recommendations proposes amending the Penal Code to create a separate crime called ``breach of duty to prevent.''
This makes it a crime if a company develops an ``AI model that performs harmful acts'' and ``fails to take reasonable measures to prevent that harm.''
The assessment also recommends "time-bound and mandatory AI safety reporting" on hacking.
This is in response to Mr Albanese's previously expressed "disappointment" that it took OpenAI months to report the hack to the Australian government.
OpenAI's rogue AI agents infiltrated Australia's Medicare health care system in June, hacking into a statistics portal containing "non-sensitive" data.
OpenAI said it first became aware of the matter in August while investigating "inconsistent activity in (AI) models" and sent an email on September 10 to the address listed on the Australian Government's services portal.
It took several more days for the email to reach the country's cybersecurity agency. It was also sent to the relevant ministers before being reported to the prime minister.
Albanese met with OpenAI CEO Sam Altman and expressed his "very serious concerns about this incident." During the meeting, Altman acknowledged there were "protocol issues" within the company.
Days after the incident was made public, OpenAI postponed the release of its new AI model, citing safety concerns.
(English article Australia considers criminal penalties for rogue AI in wake of OpenAI hack)
