The Commonwealth Ombudsman has called for laws to be changed so assessors can override decisions made by an aged care algorithm. The watchdog found the system had produced unfair outcomes for residents and weakened public confidence. Ombudsman Iain Anderson said many outcomes were inappropriate and that the department did not fully understand the legal framework it created. The government says an independent review will be released in the first half of 2027. Opposition health spokeswoman Anne Ruston said the findings showed lessons from robodebt had been ignored.
Commonwealth Ombudsman Iain Anderson has found the aged care funding algorithm is neither fair nor consistent. His report, over 100 pages, says the Department of Health, Disability and Ageing acted unreasonably in running the assessment scheme. He recommends that senior assessors be allowed to override the algorithm's decisions. The government has not accepted that recommendation, but has agreed to a review of the Aged Care Act.
A Commonwealth Ombudsman investigation has found that the government's automated aged care assessment tool, the Integrated Assessment Tool, produces unfair outcomes and erodes public trust. Ombudsman Iain Anderson said algorithmic decisions cannot currently be overridden, even when clinically inappropriate. The investigation also found the department wrongly told assessment providers that automated outcomes could be changed, and that there was no public information explaining how the tool reaches its results. The Ombudsman made 10 recommendations, including legislative changes allowing overrides.
AU Now app
Your suburb’s news, alerts and Shorts — in one app
Alerts, Shorts and your saved stories — free on Google Play.
An algorithm used by the federal government to assess older Australians' care needs is unreasonable and unfair, according to an investigation by the Commonwealth Ombudsman, Iain Anderson. The 104-page report found assessors are barred from overriding the tool's funding recommendations, even when they are clinically wrong. It also found there is no public explanation of how the algorithm reaches its outcomes and that recommendations from the robodebt royal commission on transparency were not followed. The Ombudsman says the law must change so assessors can correct clinically incorrect decisions. The Department of Health and Ageing has been contacted for comment.
Communications Minister Anika Wells has scrapped discretionary powers that would have let her define new online harms, to help pass her digital duty of care bill. The bill would require platforms to protect users under 18 from harmful content, and let users over 16 turn off algorithms. The Greens, who hold the balance of power in the upper house, will push the bill to an inquiry due to report in November. Ms Wells says she expects the bill to pass before Christmas. The Coalition said it would not support the bill in its original form.
Communications Minister Anika Wells has scrapped discretionary powers that would have let her define new online harms, to help pass her digital duty of care bill. The bill would require platforms to protect users under 18 from harmful content, and let users over 16 turn off algorithms. The Greens, who hold the balance of power in the upper house, will push the bill to an inquiry due to report in November. Ms Wells says she expects the bill to pass before Christmas. The Coalition said it would not support the bill in its original form.
Meta has introduced an AI-powered tool called "Your Algorithm" on Facebook, letting users shape what appears in their Feed and Reels. Users can describe their preferences in their own words, and changes are said to appear in recommendations in real time. The platform's three-dot menu has also been updated, with options to mark posts as "Interested" or "Not interested". Meta says Facebook may ask follow-up questions to refine its understanding of user tastes. The feature follows a similar launch on Instagram and mirrors Threads' "Dear Algo" tool, reflecting a broader move towards giving people more say over recommendations.
