A government mandate to appoint chief AI officers by July next year appears set to be fulfilled by incorporating the responsibilities into an existing senior role, rather than by hiring standalone CAIOs.
iTnews contacted 30 key federal departments and agencies to get some preliminary indications of their plans to meet Public Service Minister Senator Katy Gallagher’s directive to appoint the officers.
Gallagher, whose other main portfolio responsibility is Finance Minister, announced last month at the launch of the federal government’s whole-of-government AI plan.
At the time, Gallagher said that the CAIOs must hold an Australian Public Service (APS) rank of at least SES level 1 and that she had given departments and agencies until July 2026 to appoint them.
The deadline is six months aways, however 14 Commonwealth entities were prepared to reveal their current plans to meet the directive’s requirements to iTnews.
Of those 14 departments and agencies:
- Seven said that they already had or currently planned to delegate the CAIO role to an existing senior executive.
- Three revealed that they were seeking to appoint a standalone CAIO.
- The remaining four said that they were still considering advice from and the department overseeing government wide execution of the policy before settling on a strategy.
Out of the balance of the remaining 16 Commonwealth entities that iTnews contacted, one – the Digital Transformation Agency – provided a link to its generic policy for responsible use of AI in government but otherwise did not respond to the question.
Two agencies, the Australian Taxation Office and Home Affairs, referred iTnews to a whole-of-government statement Finance requested in lieu of direct responses from each department and agency.
The remaining 13 agencies did not respond in any form in time for publication.
Gallagher’s AI political problem
The leadership structure that Gallagher has settled on for AI oversight within departments and agencies neatly embodies an attempt to solve a political problem that AI presents for Labor.
Labor has inherited a Commonwealth public service that is falling behind the private sector on delivering service and productivity scores.
The 2019 Thodey review of the APS was the political highwater mark for the issue and now AI threatens to widen the gap it identified.
Gallagher set up a new oversight body within Finance called AI Delivery and Enablement, or AIDE.
AIDE summed up the purpose of the newly-announced CAIOs in advice that it sent to federal departments and agencies this way: “Without dedicated leadership to capture the potential of this technology, the APS is at risk of lagging the private sector and international peers, succumbing to bureaucratic inertia, and becoming exposed to harms arising out of uncontrolled AI use.”
Gallagher’s new AI commissars are part of a mission to stop that happening.
However, Labor has to balance that goal against its competing priority to ensure that AI doesn’t harm the public service and Australians broadly more than it helps them.
CAIOs versus AOs
The CAIOs appear to be Labor’s attempt to find that balance, but it places them in direct conflict AI accountable officer (AO).
The AOs, which the Digital Transformation Agency revealed mid-2024, were designed to temper unbridled take-up of AI across the Commonwealth.
Now, it would appear that they could be being set up to wrangle Gallagher’s new AI chieftains.
AIDE acknowledged the potential conflict in guidance for appointing CAIOs it has given to government departments and agencies.
iTnews has seen the guidance document and in its AIDE writes that CAIOs “will need to be willing and able to contest default processes and assumptions”.
Some might also take the view that the rhetoric that AIDE is uses in the document to grapple with the potential for conflict in Labor’s AI leadership structure is mildly Orwellian.
In bullet-points under the title “Why this works”, the document states:
- “Pure acceleration without risk management is reckless”.
- Then, immediately after in the next bullet point, it states: “Pure risk management without acceleration means we stagnate”.
Furthermore, in the same document it advises:
“Some agencies, such as smaller organisations, may opt to have both the CAIO and accountable AO roles fulfilled by the same leader, but the preference otherwise is for these to be separate.”
“Where agencies choose to appoint a single leader for both, it is recommended that the expectations and strategies for how the potential tensions will be managed are made explicitly clear,” it adds in later in the document.
Managing the tensions
This might prove to be a thorny problem for the government as departments and agencies move to execute Gallagher’s directive.
In two cases, that of the Australian Federal Police and the Australian Securities and Investment Commission (ASIC), the AO and CAIO responsibilities are to be placed in the hands of a single executive.
It’s currently not known how many of the 16 of the agencies that iTnews contacted, that did not provide a response, will do the same.
AIDE also flagged another area of potential concern in the way that Gallagher’s CAIO directive might manifest in practice.
It appears that part of the government’s motive for wanting separate chief AI officers is the government’s concern that CIOs will take an overly technocratic view of implementing the technology.
As AIDE put it in bureaucrat-speak, it’s the “risk of tech-first mentality, potentially overlooking human, cultural, or policy dimensions of AI adoption.”
iTnews found that a number departments and agencies have already given CAIO responsibilities to an executive currently in a senior technology focused role or reports to the chief information officer (CIO). More still are currently planning to do.
The Department of Foreign Affairs and Trade (DFAT) is an illustrative case. A spokesperson for the department told iTnews that not only was its CIO Andrew Smorhun its AO, but also that its CAIO, who is supposed to challenge his calls, is one of his departmental subordinates.
“The chief information officer (CIO) is currently the accountable officer (AO). The chief artificial intelligence officer role has been assigned to an SES1 officer who reports to the CIO within the Information Management and Technology Division,” a DFAT spokesperson said.
The AFP’s Manager Technology Data & Strategy, Benjamin Lamont, who the law enforcement body described to iTnews in statements as “effectively CIO”, is also the agency’s CAIO.
ASIO did not respond directly to iTnews’ questions but referred to its publicly available AI policy development document which states of its policy development apparatus for deploying the technology:
“This includes a structured decision-making framework, where accountability and decision-making are vested in a senior delegate, and informed by input and advice from internal stakeholders with diverse expertise and perspectives, who are well-positioned to assess the implications of our AI use.”
The intelligence gathering body did not provide any further comment.
ASIC’s senior executive leader data analytics & AI Graham Jefferson is the currently the securities regulator’s AO and now appears set to be the CAIO as well.
“Graham Jefferson has held the role of Senior Executive Leader Data Analytics & AI since July 2024 as part of ASIC’s vision is to become a leading digitally enabled and data informed regulator by 2030. Graham is also ASIC’s Accountable Official for the purposes of the Government’s Policy for responsible use of AI in government,” a spokesperson for ASIC said.
The Fair Work Commission said it would wait for guidance before deciding how it would handle the directive.
However, it told iTnews “given our organisational context, structure and size it is unlikely the Commission would establish a dedicated role”.
It’s not clear whether Fair Work will look to its technology division or another part of the organisation to appoint a CAIO.
The Australian National Audit Office (ANAO) currently delegates AO responsibilities jointly to its chief operating officer and chief information officers.
An ANAO spokesperson said that the agency had “nothing further to say” on Gallagher’s announcement.
The Office of the Commonwealth Ombudsman, according to it AI transparency declaration, has taken the same approach as the ANAO.
“The chief operating officer and chief information officer are the designated AI Accountable Officials,” it wrote in its statement.
“We will work towards the appointment of a chief AI officer as per the whole-of-government AI plan in accordance with the July 2026 deadline,” an ANAO spokesperson said.
Finance’s committee approach
Perhaps fittingly for a large department, Finance delegates its AO responsibilities to an entire committee. It’s co-chaired by two AOs with multiple senior executives across divisions that contribute to the committee’s activities and deliberations.
“The Department of Finance has an AI Governance Committee to oversee the adoption and use of AI within Finance.
“The committee is co-chaired by two accountable officials (AO) and its membership includes a number of senior executive representatives from across Finance,” a spokesperson for the department said.
The spokesperson said that Finance expected to appoint its CAIO early next year.
Austrade, the Department of Climate Change, Energy, the Environment and Water, the Office of the Australian Information Commissioner, and the Department of Industry Science and Resources are still considering their options.