AI body suddenly scrapped after 15 months spent finding experts
Tim Ayres (left) and Andrew Charlton announced plans for an AI safety institute in December. (ABC News: Matt Roberts)
In short:
A permanent AI advisory body was scrapped after the government had already spent 15 months working to identify experts to sit on it.
Experts fear ongoing delays to AI regulation could mean the government misses a limited chance to properly safeguard against artificial intelligence harms.
What's next?
A replacement AI safety institute will be established in early 2026.
It took 15 months to narrow a field of 270 artificial intelligence experts to 12 nominees for an AI authority, at a cost of almost $200,000, before it was suddenly scrapped by the federal government.
The AI advisory body was promised by former industry minister Ed Husic in 2024 after he appointed an interim expert panel to begin work on "AI guardrails" that could ensure the technology was safe.
The body was funded in the 2024 Budget in a $21.6 million package that also included an expansion of the National AI Centre.
But emails between the office of Industry Minister Tim Ayres, who replaced Mr Husic, and his department reveal the promised body was scrapped just months after the department had spent $188,000 and more than a year whittling down a list of experts to a shortlist of 12 nominees.
The shortlist of nominees was contacted by Mr Husic in February 2025 to collect the required "documentation" for their appointments, but they heard little between then and August, six months later, when the body was formally scrapped.
Senator Ayres and Assistant Technology Minister Andrew Charlton announced in December that the federal government would instead establish an AI safety institute, which like its predecessor would sit within the department, at a cost of $29.9 million.
The government says the institute will be established early this year.
In a statement, a spokesperson for the minister said the new institute would have the capability to test, monitor and advise on gaps in regulations "if and as they emerge", rather than relying "solely" on external expertise.
"Previous proposals for an AI Advisory Board have been superseded by a more dynamic and responsive approach to keeping Australians safe as AI is adopted across the economy," the spokesperson said.
"The Department of Industry, Science and Resources and its ministers are continually engaging with a wide range of experts as they deliver that approach."
Advice prepped to explain pivot at 15-month mark
The original body was abandoned as part of a pivot by the federal government away from Mr Husic's "mandatory guardrails" for artificial intelligence, which were anticipated to require legislation and even a possible standalone AI Act, in favour of a lighter-touch approach to the growing technology.
In an email to the department on August 7, an adviser to Senator Ayres wrote they had sent a brief on the AI advisory body the day before, and "just wanted to check when you'll be going back to the potential candidates and informing them of the minister's decision".
"Would be good to coordinate the response to any potential concerns," the staffer wrote.
The department responded it would inform the 12 nominees the body would not be proceeding and ask its contractor to advise all other applicants, though the minister's office ultimately directed the department not to inform the list of 264 applicants.
Anticipating a backlash, the minister's office wrote in prepared talking points "if asked" why the advisory body was no longer going ahead that the "rapidly evolving landscape of AI" required a rethink of how the government accessed expert advice.
"The government is engaging widely and accessing expertise through processes like the economic reform roundtables," the talking points said.
"It is also considering submissions received through a range of processes including the Productivity Commission's productivity inquiries."
Officials were advised to respond "if pressed" on "why it has taken 15 months to come to this decision" that the process to identify candidates had been rigorous and comprehensive.
The Productivity Commission last year urged the plan for "mandatory guardrails" on AI be put on ice, or else it would risk throttling AI's $116 billion economic potential.
Professor Toby Walsh, who was part of the AI expert group, has expressed his fears Australia is missing a narrow window of opportunity to regulate AI. (ABC News: John Gunn)
The government said while it had spent $188,000 and 15 months on advertising and recruiting for the AI advisory body, the "insights gathered" through that process would be valuable for shaping future AI policy.
After the expert body was scrapped Professor Toby Walsh, one of its members, expressed his concern that the federal government was missing a narrow window of opportunity to ensure the safe implementation of AI.
"I'm just nervous we are going to repeat the same mistakes [as with social media], possibly on steroids," he told ABC Radio National last year.
"I am just fearful that in a couple of years time I will be on your show again talking about how we missed the opportunity to pre-emptively prevent some of these [harms] happening."
The assistant minister, Mr Charlton, said in December the new AI safety institute would be able to "tap into" expertise, whether from overseas counterparts, tech experts or unions and business.
Mr Charlton explained in a speech last year the government had decided the best way to keep Australians safe was not to "create a single agency responsible for AI" but to "empower every existing agency across government to take responsibility for AI".
"Instead of creating a new source of authority, we want every existing agency to be responsible and empowered to address the risks and opportunities of AI," he said at the time.
The government says the new institute will be able to advise on strengthening or updating regulations and legislation "as needed".