decrypt.co
The guidelines, released late Wednesday, emphasize risk management, transparency, human oversight, and fairness to ensure AI systems operate safely and equitably.
While not legally binding, the country’s standards are modeled on international frameworks, particularly those in the EU, and are expected to guide future policy.
Dean Lacheca, VP analyst at Gartner, acknowledged the standards as a positive step but warned of challenges in compliance.
“The voluntary AI safety standard is a good first step towards giving both government agencies and other industry sectors some certainty around the safe use of AI,” Lacheca told Decrypt.
“The…guardrails are all good best practices for organizations looking to expand their use of AI. But the effort and skills required to adopt these guardrails should not be underestimated.”
The standards call for risk assessment processes to identify and mitigate potential hazards in AI systems and ensure transparency in how AI models operate.
Human oversight is stressed to prevent over-reliance on automated systems, and fairness is a key focus, urging developers to avoid biases, particularly in areas like employment and healthcare.
The report notes that inconsistent approaches across Australia have created confusion for organizations.
“While there are examples of good practice throughout Australia, approaches are inconsistent,” the government’s report notes.
“This is causing confusion for organizations and making it difficult for them to understand what they need to do to develop and use AI in a safe and responsible way,” it reads.
The framework highlights non-discrimination, urging developers to ensure AI does not perpetuate biases, particularly in sensitive areas like employment or healthcare.
Privacy protection is also a key focus, requiring personal data used in AI systems to be handled in compliance with Australian privacy laws, safeguarding individual rights.
Additionally, robust security measures are mandated to defend AI systems from unauthorized access and potential misuse.
Edited by Sebastian Sinclair