News
Officials’ concern over AI use in Medicare billing
An Audit Office report into the Health Department’s use of AI in compliance cites prompted billing software as ‘an emerging issue’.
The RACGP has called for caution in the use of AI to assess Medicare integrity risks.
An Australian National Audit Office (ANAO) assessment of AI in Medicare compliance published on Monday has revealed official unease about the adoption of prompted billing software.
Its report, ‘Artificial Intelligence and Medicare Benefits Integrity’, found Services Australia had expressed concerns ‘about providers using AI to test MBS eligibility for multiple items and suggest billing options’.
These were raised with the Department of Health, Disability and Ageing (DoHDA), which has ‘not yet systematically identified, assessed and established treatments for AI risks’, according to the ANAO.
The auditor found an internal intelligence report drafted this May by DoHDA’s Benefits Integrity Division pinpointed prompted billing software as ‘an emerging issue,’ and identified four commercial tools currently used to analyse consultation notes.
This report revealed that officials are currently unable to identify which practices or healthcare professionals are using the AI-based functionality but suggested ‘most inappropriate claiming should be detectable through existing capability’.
While the Department’s intelligence report put forward initiatives to identify the use of this type of software, the ANAO notes it had not been finalised as of last month.
For Dr Tony Bayliss, Chair of the RACGP’s Regulation and Compliance Working Group, the ANAO findings reflect the importance of human oversight both within general practices and official compliance work.
‘It reaffirms the fact that GPs are ultimately responsible for the item numbers that are in their billing,’ he told newsGP.
‘That is not a responsibility that can be outsourced to AI, in much the same way that we should be cautious about outsourcing AI to determine irregularities in billings as well.
‘It’s advice, but it should not be taken blindly.
‘Ultimately with AI, you do need to be aware of the possibility of hallucinations, and certainly the medico-legal risk of hallucinations in your billings would be significant.’
More broadly, the ANAO found the Department’s management of AI in supporting Medicare integrity to be ‘partly effective’, although advised it ‘could be more proactive’.
Overall use of AI was described as ‘limited’, but the ANAO said DoHDA used an AI-enabled model between July 2024 and December 2025. It was at least partly involved in identifying eight potential fraud or non-compliance cases costing around $5.2 million.
For Dr Bayliss, this previous use of AI compliance is a concern without engaging with the college.
‘Any project of this nature requires goodwill and not involving the medical community and GPs decreases the goodwill there is to support a change of this magnitude,’ he said.
‘What we need is engagement and transparency in the process.’
He also noted that overall non-compliance is a ‘relatively small proportion’ of Medicare billings, ‘especially given that we understand that the bulk of this is unintentional’.
‘This represents the complications of the Medicare system and in particular the item numbers,’ he said.
‘So using AI to detect irregularities in billings misses the point that it’s the system complexity that is the cause for the bulk of the incorrect billings.’
He also warned of the potential for ‘significant unintended consequences’, including diverting resources to compliance activity when appropriate billings have been incorrectly flagged by AI.
The ANAO report makes three recommendations to DoHDA:
- Work with other entities to assess AI risks, including its use in Medicare claiming
- Strengthen compliance
- Strengthen assurance, including by updating cyber security policies that ‘reflect the unique risks posed by artificial intelligence’
In its response, which is included in the ANAO report, DoHDA said it ‘recognises that artificial intelligence presents both opportunities and emerging risks across the health system and government’.
‘As artificial intelligence adoption has increased, the Department has strengthened its governance, risk management, assurance and oversight arrangements and will continue to mature these in line with evolving whole-of-government requirements and sector practices,’ it said.
It agreed with all the ANAO recommendations and said it is already carrying out work in several of the areas identified by the auditor.
The RACGP had voiced concerns earlier this year that the use of AI by DoHDA for Medicare compliance could lead to GPs being unnecessarily targeted.
In
a submission to the ANAO in April, the college called for restraint in its use given the risks.
‘If used, it should be recognised that there is a spectrum of compliance reflecting different practice arrangements, and it is dangerous to rely solely on arbitrary standards set by Government,’ the RACGP stated.
‘This audit also specifically refers to “non-compliance”, which neglects the opportunities presented by AI to identify good clinical practice, reduce red tape and improve workflows in general practice.’
Log in below to join the conversation.
AI ANAO MBS billing Medicare Integrity
newsGP weekly poll
Have you ever found out that a patient has previously failed to disclose their use of GLP-1 medicines to you as their usual GP?