Victorian premier defends Labor’s AI ad making opponent voice invented health-cost warnings

Victorian Premier Ben Carroll defended Labor’s AI-generated attack video on November 4, arguing it highlighted reasons to fear the opposition’s health plan. But the video does more than criticize policy: it depicts Liberal leader Jess Wilson delivering healthcare-cost warnings she never uttered. Wilson called it a “desperate smear campaign” and rejected its portrayal of her plans.

A policy answer becomes a synthetic speech

Posted to Victorian Labor’s social accounts on Halloween, the video shows a generated Wilson trick-or-treating beside Warren Pickering, One Nation’s Victorian leader. Pickering wears a Liberal badge. The artificial Wilson speaks in a voice resembling hers and holds a bill listing supposed US-style healthcare charges.

Those figures include $150,000 for heart-attack care, $6,000 for an ambulance and $40,000 to have a baby. They are claims presented by the fabricated character, not charges Wilson announced or words she actually spoke, the Guardian reported.

The attack draws on an exchange at a press conference the previous month. Wilson had announced a $1.27 billion plan to pay private hospitals to perform public elective surgeries, aiming to reduce the waiting list. A reporter asked whether it was bad to “Americanise” the state’s healthcare system.

Wilson answered, “Not at all.” She then said the arrangement was happening in Queensland and had previously been implemented by Labor in Victoria. That explanation concerned using private hospitals for public surgery; the AI performance supplies a separate speech about patients facing large bills.

Carroll defends the political argument

Speaking to Raf Epstein on ABC Radio Melbourne, Carroll said the video highlighted why Victorians should be “terrified” by the Coalition’s health plan. His defence rested on Wilson’s earlier answer: “Jess Wilson herself has said there is nothing wrong with Americanising our health system.”

Wilson rejected that interpretation, saying that “in no way, shape or form are we Americanising the healthcare system here in Victoria”. The Liberal campaign dismissed the video as “garbage” and urged voters to treat it that way.

University of Sydney media researcher Timothy Koskie described a risk beyond viewers accepting a fabrication literally. Artificial images and videos can make subtle mischaracterisations more convincing, he told the Guardian. He also argued that parties have little incentive to hold back because the public will not reward restraint.

Trying to clean it up after is a losing proposition

Timothy Koskie, University of Sydney media researcher, speaking to the Guardian

Labels, consent and campaign commitments

Clancy Moore, chief executive of Transparency International Australia, warned that fabricated politicians and deceptive ads could further erode political trust. He called for Victoria and the federal government to introduce truth-in-political-advertising laws. At a minimum, he said, Victorian parties should use an AI watermark to inform voters.

South Australia provides the model cited by reform advocates. Its rules ban AI-generated electoral deepfakes depicting someone doing something they did not do without their permission. A person can consent to such use of their image, but that advertisement—and other advertisements using AI—must still be labelled.

Victorian Greens leader Ellen Sandell said the party would call on whoever forms government to introduce South Australia-style laws. She also urged parties to sign a campaign memorandum setting standards of behaviour, including a commitment not to use “deceptive AI content”. That seeks a campaign commitment alongside the proposed legislation.