Scare campaigns: Victoria’s major parties trade blows over AI-generated videos of leaders in election ads

Political parties in Victoria are engaging in a campaign of scare tactics through the use of AI-generated videos featuring leaders, raising concerns about the erosion of trust in politics. A recent video posted by Victorian Labor depicts Jess Wilson, the Victorian Liberal leader, warning of the costs of "Americanised healthcare" under a Liberal government.

However, Wilson never made such a claim, and the video's content stems from her remarks at a press conference last month. Ben Carroll, the Premier of Victoria, defended the video's content, stating that Wilson herself had said nothing wrong with Americanising the state's healthcare system.

The Liberal Party has criticized the video as a "desperate smear campaign" and dismissed it as "garbage." However, the party itself is also using AI-generated content on social media, including videos of Premier Ben Carroll asleep at his desk. These tactics come on the heels of allegations by unnamed sources that Labor leader Daniel Andrews takes naps on the job.

The Conservatives have even released an AI-generated video depicting a Labor victory in the election, complete with former premiers Jacinta Allan and Daniel Andrews.

Experts and political parties, such as the Greens, are calling for the ban of AI-generated videos in election ads to protect public trust in politicians and democracy. Tim Koskie, a media researcher at the University of Sydney, acknowledges that misinformation in political messaging predates AI, but the rise of fake videos and images makes it easier to create more convincing arguments that don't hold up.

He suggests that those spreading misinformation should be held accountable for their mischaracterizations, with even small fines being effective in curbing bad behavior.

Moore, the chief executive of Transparency International Australia, is calling for the introduction of truth in political advertising laws in Victoria and the federal government. Currently, these laws exist in the ACT and South Australia, banning the use of AI-generated deepfakes to depict acts a person did not perform without their permission.

A person can consent to their image being used in a deepfake advertisement, but those advertisements and any others that use AI must be labeled as such. Moore also suggests the use of AI watermarks to inform voters.