New laws aimed at outlawing the use of artificial intelligence to generate violent and sexually explicit deepfake images, audio or videos will be introduced to State Parliament this week.
Under the proposed laws, using AI to generate wholly simulated humiliating, degrading or invasive images, audio or videos purporting to be of real people would be outlawed with offenders facing up to four years in prison or fines of up to $20,000.
The laws would cover images, audio and videos generated by AI that include acts of assault, violence or cases where a simulated person –purporting to be a real person– is either in a state of undress or performing acts that fall outside the standards of morality and decency that would be accepted by reasonable members of the community.
Authorities estimate as much as 90-95 percent of deepfakes are non-consensual porn; and 99 per cent of victims in deepfake porn are women.
The eSafety Commissioner has stated explicit deepfakes have increased on the internet as much as 550 per cent year on year since 2019.
The Government thanks Upper House MLC the Hon. Connie Bonaros for her advocacy on this important matter and looks forward to continued collaboration with Ms Bonaros on creating legislation to protect those targeted by violent and sexually explicit deepfake images, audio or videos. Labor member for Florey, Michael Brown MP, has also done extensive work on this issue as Chair of the Select Committee on Artificial Intelligence and was instrumental in the development of this Bill.
Following the release of a discussion paper in August this year, the Government is separately giving consideration to the broader issue of deepfakes beyond the parameters of non-consensual sexually explicit material, such as misuse in scams, blackmail, political misinformation and fraud.
As put by Kyam Maher
Technology that relies on artificial intelligence is evolving at a rapid rate, and the extent to which AI can be used to generate entirely fake videos and images is utterly chilling.
While we are continuing to consider legislative reforms to tackle deepfakes, we are also pursuing this avenue to ensure any vile, fake material that has been wholly generated by AI is outlawed in South Australia
In addition to the tough fines, we're also giving courts the power to seize any materials used in the commission of an offence, including computers, mobile phones or other hardware employed by an offender to create or distribute this material.
This legislation sends a clear message that South Australia will not tolerate the use of AI to develop sick, degrading material.
As put by Michael Brown MP
The Select Committee highlighted some of the risks associated with AI technology, including the worrying trend of the increase in image-based abuse.
Whether it's material of a real person that has been edited, or something entirely fake that purports to be of a real person, the fact is any material of this nature should be unlawful.
These changes will ensure material generated wholly by artificial intelligence are dealt with appropriately.
As put by Connie Bonaros MLC
AI has made humiliation possible in an instant. Territory that was only in the realm of hackers and underground users a year ago is now available to all of us - literally at the click of a button. The consequences for kids and adults alike, especially when sexually explicit material is created, can be absolutely devastating.
As a parliament, we share a collective responsibility to respond swiftly to emerging technologies and to those who use these technologies for malicious and harmful purposes.
