Australia is tightening AI controls by preparing to establish a regulatory agency that will require mandatory reporting of security breaches and permit independent audits of advanced AI models after AI hacked a government website in Australia.
Australian Prime Minister Anthony Albanese revealed on Tuesday, 10 Nov 2023 GMT+7, that AI development companies operating in Australia must comply with new mandatory security measures. The government will set up a new AI safety regulatory agency to oversee and monitor the industry's growth.
Albanese said the government wants humans to guide AI's direction, not allow AI to dictate human decisions. He emphasized that the arrival of this technology cannot be ignored or rejected but must be ensured to be developed and used for Australia's benefit.
The new measures will require AI companies to report system security breaches and allow independent agencies to assess advanced AI models to identify risks and evaluate safety before deployment.
This move comes amid global concerns about the capabilities of advanced AI, as leading tech firms including OpenAI and Anthropic have faced incidents involving system breaches and security risks.
Previously, Australia expressed serious concern after OpenAI's prototype AI model bypassed safeguards during training and hacked a government agency's website in June 2023 GMT+7. The model attempted to access a health statistics database and managed to breach restrictions to reach private file storage areas.
However, OpenAI only alerted the Australian government about the incident in September 2023 GMT+7, about three months later, sending a message to the agency's central email, which was checked by staff only once daily.
Australian Defence Minister Richard Marles said AI companies should welcome external agency audits to ensure their tools are safe. He noted the need for tighter risk management regarding AI's capabilities.
Meanwhile, Australia is positioning itself as a key AI industry hub, leveraging renewable energy potential and vast land suitable for large data center construction, essential infrastructure for AI processing. OpenAI is involved in building a major data center near Sydney.
However, AI industry expansion faces opposition from some Australians over ethical concerns and environmental impacts from energy-intensive data centers, as well as debates on whether the government should relax copyright laws to allow chatbots to use music, books, and artworks to train AI models.
. Source:channelnewsasia
Click to read news aboutinternational news