Legislation covers images generated entirely via digital technology
The South Australia Attorney-General’s Department has announced that its new laws aiming to address the use of artificial intelligence (AI) to create violent or sexually degrading deepfakes, which can be images or videos, have taken effect as of Nov. 3.
The new laws – dubbed in the attorney-general’s news release as tackling the ‘dark side’ of AI – passed the state’s Parliament earlier this year.
Under the new laws, those who utilize AI or other digital technology solely to generate invasive or humiliating images or videos that closely resemble or purport to be real individuals may face fines of up to $20,000 or four years in prison and may need to surrender the records and equipment used to create the deepfakes.
In the news release, the attorney-general noted that the Summary Offences Act 1953 (SA) has image-based offences targeting deepfakes created by editing or manipulating actual images. However, the attorney-general highlighted that the new laws will cover deepfakes generated entirely by AI or other digital technology.
The attorney-general’s news release shared the following data:
Michael Brown, SA’s assistant minister for AI and chair of the Select Committee on AI, devoted extensive efforts to this problem.
Reports noted that deepfakes in the country and beyond have targeted students.
Under SA’s child safety curriculum reforms to address this issue, the state’s schools now have an evidence-based program aimed at preventing or mitigating problems arising from or associated with social media, online safety, cyberbullying, body image, mental health, and wellbeing.
The curriculum currently covers coercive control, deepfake detection strategies, AI comprehension, and the effects of negative online influencers.
“These reforms are a great example of how we need to not only ensure our children are educated on the fast-evolving changes of technology, but also ensure our laws are up-to-date to afford that legal protection from online harms,” the attorney-general’s news release said.
Relatedly, on Oct. 13, the attorney-general announced that laws prohibiting the use of election-related robocalls, robopolls, and deepfake political advertisements took effect in advance of the March 2026 state elections.
The electoral reforms banned the unauthorized use of AI-generated deepfakes to make electoral ads showing acts an individual did not carry out, with offenders facing maximum penalties of $5,000.
“A person will be able to consent to their image being used in a deepfake advertisement, but those advertisements - and any others that use AI - will need to be clearly labelled as such,” the attorney-general’s prior news release said.
“The laws prohibit using AI to generate electoral advertisements, or deepfakes, which have been designed to mislead voters, which has occurred in recent interstate and overseas elections,” the news release added.
