Dispute over medical costs
On Halloween night, a video appeared in social media feeds showing Jess Wilson warning of the “Americanisation of healthcare” under a future Liberal government. In the video, she holds a bill listing supposed US treatment costs: $150,000 for a heart attack, $6,000 for an ambulance and $40,000 for childbirth. Warren Pickering, leader of One Nation’s Victorian branch, stands beside her wearing a Liberal Party pin.
Wilson never said those words: her image and voice were generated by AI. The video was posted on Saturday on Victorian Labor’s social media accounts.
The figures in the video relate to a press conference last month. Wilson announced a $1.27 billion plan to pay private hospitals to perform elective surgeries for public patients, in an effort to reduce waiting lists. Asked whether “Americanising” the state’s healthcare system was a bad thing, she said no, adding that a similar arrangement was already in place in Queensland and had been used by Victoria’s Labor government in the past.
On Wednesday, Premier Ben Carroll defended the video’s message on ABC Radio Melbourne. He said Wilson herself had said there was nothing wrong with Americanising the system, and added that Victorians should be concerned about it. Wilson called the video a desperate smear campaign and said Victoria was not moving towards an American healthcare system in any sense. The Liberals called the video “rubbish” and urged voters to treat it accordingly.
But the party itself uses AI in its election material. In one video, Premier Ben Carroll falls asleep at his desk. It appeared after an October 24 front-page story in the Herald Sun, in which unnamed sources accused the Labor leader of dozing on the job. A source from Labor’s campaign team, who was not authorised to speak publicly, said the Liberals had “thrown the first stone” by making the video.
Other political players are also releasing AI videos. Last month, the conservative lobby group Advance Australia circulated a video depicting a fictional Labor election-night victory. It featured former premier Jacinta Allan and Daniel Andrews wearing a North Face jacket. At a One Nation campaign launch, a video was shown featuring provocative stereotypes about people from different ethnic groups.
How AI is changing political disinformation
Clancy Moore, head of Transparency International Australia, says trust in politicians is already at a record low. In his assessment, lies, deepfakes and AI-generated images of politicians in election advertising further undermine trust in democracy in Victoria.
Timothy Koskie, a media researcher at the University of Sydney, notes that disinformation in political posts predates AI. But fake videos and images make it easier to spread distorted versions of events. The goal of such material may not be to persuade voters of a direct lie, he says, but to make a case that does not withstand scrutiny seem more convincing. Stopping the practice is difficult: Koskie believes voters will not reward those who give up these tools.
Koskie calls for those spreading disinformation to be held accountable for distorting facts. In his view, correcting the damage after publication is not very effective. He says even small fines can deter violations, citing X’s $650,000 fine for failing to comply with an Australian child-safety requirement. Small measures, he adds, can produce noticeable results over time: a mistaken claim does not have to lead to its author being publicly punished, but it should have consequences.
Moore is urging the Victorian and federal governments to pass laws requiring political advertising to be truthful. Such rules are already in force in the Australian Capital Territory and South Australia.
In South Australia, it is illegal to use an AI-generated deepfake in election advertising without a person’s consent if it depicts them doing something they did not do. Violations can incur fines of up to $5,000. A person can give permission for their image to be used in a deepfake, but such advertising, like other material made using AI, must be labelled.
Moore says Victorian political parties should at least watermark AI-generated content. Of the videos reviewed, only the Advance Australia video carried a warning: “This nightmare election-night scenario was created with AI.”
On Thursday, the Greens plan to call on whichever party forms government to adopt South Australia’s rules on truthful political advertising. State party leader Ellen Sandell said Victorians should know what they are seeing and hearing from political parties. She also proposed that parties sign a memorandum on campaign conduct standards, including a commitment to avoid “deceptive AI-generated content”.
Dataist’s take: The dispute in Victoria shows the limits of voluntary warnings: of the videos reviewed, only Advance Australia added a label. South Australia’s rules provide a concrete mechanism — consent, mandatory labelling and fines — and it is this model that is being proposed for Victoria and the federal level.
Daily AI news
Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.
Only what matters — every day
Follow on X