This post is the final one for the theme the first two have been exploring. I previously explored whether sufficient laws were satisfactory for manslaughter, why anti-hate speech laws were needed to handle Nazi vilification and in this one I want to make the claim that new laws will need to be introduced to handle Generative Artificial Intelligence (GenAI) use to generate images of a criminal nature.

GenAI is a relatively new phenomenon occurring within the last few years to the degree that it is widely known as at the time of writing in late 2026. GenAI has the potential to allow human expression in new ways, but it is not this articles point or contention to explore the controversies and issues surrounding how AI models are trained and what they mean for artists. Rather I want to explore whether existing laws are able to handle the influx of AI and the likely demand for governmental response as the mainstream public becomes aware of how prominent generated images become.

Three areas emerge for GenAI to cause a public outrage, and the public demands a response. The first is the creation and distribution of Child Sexual Abuse Material (CSAM), Deepfakes and Revenge Porn. All three of these are heavily criminalised, but I want to better explore whether these existing laws are satisfactory in handling GenAI and the potential for great change contained within.

For this article I wanted to talk to someone within the field of Information Technology (IT) and who has some experience with GenAI and other AI through education and experience. However, after repeated weeks of setting up agreed meetings and them not responding, I was unable to have them commit to this article and will have to continue the best I can.

Child Sexual Abuse Material

CSAM is a broad term that includes a spectrum of material including sexualised videos and images to children being exposed to sexual content. Section 51A of the Crimes Act includes a definition of child abuse material, it includes a child or implied child as a victim of torture (including sexual), sexual abuse or engaged in a sexual activity or pose, showing their genitalia or breast area (if female) or something a reasonable person would consider offensive.[1] A similar definition is seen under the Crimes Legislation Amendment (Telecommunications Offences And Other Measures) Act (No. 2) as it provides a definition largely similar to the previous definition as it focuses on depictions of a person who is under 18 or appears to be engaging in a sexual pose or activity, in the presence of someone doing a sexual pose or activity, something objectively offensive, or the dominant characteristic is sexual (sexual organ or anal region, female breasts) or implied sexual activity or pose or material describing the aforementioned things. ^[2] This includes videos and images of children in sexual poses, being raped (as they cannot consent no matter the circumstance), and audio recordings of children sexually. This kind of material can be generated through all kinds of methods. Video recordings are one of the most common ways to generate this kind of material, as this allows for still images to be shared and disseminated.

Thankfully, this kind of material is illegal and heavily prosecuted as there are few legitimate reasons to own this kind of material. Looking to the Crimes Act s 51 it focuses on different aspects of child sexual abuse material. It is illegal to produce child abuse material,[3] to distribute said child abuse material,[4] and to possess child abuse material.[5] Furthermore, this kind of behaviour is illegal through the Commonwealth as well, as it is also illegal to possess, distribute or produce child abuse material.[6]

Through these existing laws, it is clear that existing legislation is sufficient to catch all facets of CSAM. However, one thing that is evident is the lack of explicit mention of GenAI. It is possible to claim that it is caught by production and distribution, but this is something I believe should be clarified.

GenAI and CSAM

GenAI is an issue in this area, as it allows for more CSAM to be produced, especially material to cater for different fetishes or other specific interests. It is being used to produce more CSAM at increasing rates, and this makes it harder for law enforcement agencies to properly enforce existing legislation or even devote resources to combat CSAM involving real children. A good demonstration of this issue is that in 2024 the National Centre for Missing and Exploited Children reported an increase of 1,325% of reports involving GenAI CSAM from ‘4,700 reports in 2023 to 67,000 reports in 2024’.[7] This must be considered in conjunction with what GenAI allows, at the time of writing GenAI had become much more advanced and was at the point where it was able ‘to generate hyper-realistic content, perpetrators can now produce convincing synthetic images of abuse’.[8] Gone are the days of Will Smith looking weird eating spaghetti and it being obviously fake, now GenAI has gotten smart enough to generate a convincing video of Will Smith eating spaghetti. All of this must be considered when thinking about CSAM production, it may not involve real children, but it raises serious issues about whether the aforementioned laws are sufficient to handle this.

It would’ve been of benefit to have the experience and insight of the failed collaborator as they could have explained how GenAI is trained and how it generates its content. AI does not generate new things on its own, it is best understood as from examples and linking thoughts together to achieve a goal. GenAI acts similarly, as it cannot create something from nothing, such as asking a blind person to describe a chicken; rather, it has to be taught things so that it can learn and understand them. When it comes to generated CSAM every AI image starts with photos of real children, whether sexual or innocent meaning that a child is harmed somewhere in the process.[9] This means that for GenAI tools to create AI CSAM material then they need to be trained on existing images, as without it they could not generate the child victims in the desired ways.[10]

Interestingly, AI-generated CSAM is a broad designation, and it can be divided into multiple types:

Wholly synthetic images: this is where the AI creates fictional children in sexualised scenarios, real children are not in the photo, but it can appear realistic.

Morphing or hybrid images: this is where AI blends pictures of children and adults creating new ones depicting them in sexual scenarios.

Cartoon or animated material: this is where AI creates styled depictions of minor, on a spectrum from crude sketches to realistic images.[11]

As highlighted above, existing legislation should be sufficient to prevent the production, distribution and collection of this material but as based on the rise of people creating AI CSAM and subsequently consuming it, it may lead to growing public concern and outrage; especially as the amount of people being caught with it is steadily increasing. A man in Western Sydney was caught with about 100 images of GenAI CSAM, including videos and a GIF animation.[12] Another Sydney man had a ‘significant amount of AI-generated child abuse material’;[13] another man in New South Wales had more sinister GenAI CSAM as he had material featuring him in it.[14] The material is illegal, as made evident by AFP Detective Acting Superintendent Luke Needham.[15] Two Australian men were charged with generating GenAI CSAM as part of a global operation, and that it is getting harder for police to determine whether the CSAM depicts a real child or is GenAI.[16] A man in Victoria used a GenAI tool to generate ‘793 realistic child abuse images’, this just goes to show how AI is evolving at a substantial rate per Victoria Police Detective Superintendent Tim McKinney from Cybercrime Division.[17] Two years ago in 2024, Tasmania had its first conviction related to GenAI CSAM;[18] showing that in the last couple years the rise of GenAI has become more and more serious. This raises the question of if in the coming years, as AI has become more and more prevalent within society whether public outrage over GenAI and its abilities to create CSAM more readily than ever will create demand for politicians to add new laws specifically covering GenAI CSAM.

This issue has received attention from politicians as Independent MP Kate Chaney wanted to introduce a bill criminalising the creation and distribution of AI tools used to create GenAI CSAM, the bill also seeks higher penalties for possessing GenAI CSAM using real children’s likeness.[19] Attempts have been made to amend the Criminal Code Act 1995 (Cth) intending to create new offences ‘offences targeting the misuse of emerging technologies …’, although this Bill did not pass, so it is unknown what the future holds.[20] It is accepted that existing legislation is sufficient to handle GenAI CSAM as it currently stands, but I believe that as people become more aware of it, there will be a political push to enact new laws to create new offences that carry greater punishments for creating, consuming or distributing GenAI CSAM. As current legislation does not address AI use in generating such material,[21] the output has been criminalised, the consumption and distribution has been criminalised but the potential production of GenAI CSAM is a blank area in law.

Deepfakes

Deepfakes are an interesting modern age problem, as they are both easy to create and as recently seen during elections worldwide can convince and fool a whole lot of people. One definition focuses on images, videos or sounds generated by AI of a real person edited to look real but are false depictions of them doing or saying things they did not do or say.[22] They are also able to be generated through machine learning creating a realistic but fabricated image or video including ‘virtual performances, face swapping, lip-synching or voice synthesis’.[23] I was unable to find a definition within existing Victorian legislation, but these definitions are sufficient to capture a spectrum of behaviours.

Deepfakes like GenAI CSAM require data to be fed into them so that they can create the desired output, they are unable to create something from nothing and require prompts and existing data to learn. What makes this technology dangerous and worrisome is that the images do not need to be sexual, or intimate in nature. It is possible to create deepfake sexualised images of someone through uploaded images taken online, from close friends, etc.[24]
An example of this would be people digitally implanting someone’s face onto images of naked celebrities with a similar skin colour and body shape.[25] This scenario is coined as ‘morphed porn’ as this kind of material was found on a thread calling them ‘morphs’.[26]  This technology has however, become more refined as all one needs ‘to create a realistic pornographic deepfake, all a perpetrator required was an image’.[27]
To highlight how easy it is to generate a deepfake of someone, Qoria provides an example of how a student can generate a deepfake of their teacher; the student needs to download an AI-powered deepfake tool to manipulate images or videos of staff; then the student looks for a publicly available image or video of the staff, which is easily doctorable; then the student uses tools to ‘swap faces, superimpose images’ or create new content; and share the manufactured content through online platforms.[28]
These two examples show how fundamental an issue that deepfakes have become within society and lead to a fundamental issue of whether existing legislation is sufficient to handle deepfakes, and whether it accounts for growing and developing kinds, as new technology and methods become apparent.

Deepfakes and existing legislation

Existing legislation focuses on different parts of the issue, some treat it as sharing non-consensual images and others treat it as offensive material; all however, fail to create a uniformed view of deepfakes and ways to handle them.

Deepfakes may fall under three categories of offences; ‘harassment, IBA and obscenity/indecency’.[29]

Under Commonwealth law there it is a crime to use a carriage service to menace, harass or cause offence, this would catch people sharing the deepfake material as people would consider it to be menacing, harassing or offensive.[30] Furthermore, it is also a crime to use a carriage service to transmit sexual material without consent, this would mean that nude deepfakes should be caught under this charge, as they are private sexual material,[31] although it is hard to say how applicable they are in practice.

However, under Victorian law there is no general harassment charge, but it is possible for sharing these deepfakes to be caught under stalking legislation. Stalking legislation is quite broad and includes actions designed to cause physical or mental harm, arouse apprehension and fear in the victim or others.[32] Interestingly, stalking has been applied to times where people have posted photographs of the victim to pornographic websites,[33] and should apply to deepfakes being posted.[34] These may not apply in situations where the perpetrator has not posted enough material, as only posting one image may not be sufficiently caught under stalking legislation, due there not being a continuous course of conduct, and that there needs to be a relationship between victim and perpetrator.[35]

Victorian Legislation already criminalises actions where intimate images are intentionally distributed where the image would offend community standards,[36] and threats to distribute intimate images.[37] However, these may not be effective in handing the distribution of deepfake images, as factors relevant to community standards may not be caught.[38] An example of this is the circumstances where the image was captured, as deepfake images are created and not captured.[39] Similarly, it is an offence to use an ‘on-line information service’ to publish, transmit or make available ‘objectionable material’.[40] This is quite vague, and is likely to cover all kinds of materials, from CSAM to Hate Speech material, gore and sexualised deepfake images. However, a sexual image created without consent depicting sexual activities likely to offend a reasonable adult,[41] should catch bestiality, CSAM and other perverted sexual activities. However, like the previous sets of legislation it suffers a similar fatal flaw, and it requires some ‘level of distribution or publication’, and that someone creating these deepfakes for personal use may not be caught under these specific offences.[42]

New offences may need to be created to specifically catch deepfake content, as existing legislation is lacking. Distribution is hard to apply for all cases, as it requires either a connection, frequency or sharing. Furthermore, creation of these images for personal consumption is largely uncaught, as long as the person does not create already illegal content for personal use.

Deepfakes ‘are a serious threat facing society’,[43] as they allow users to create all kinds of degenerate content for their own enjoyment. CFake a deepfake website allowed deepfake material of ‘famous women, including politicians, first ladies of multiple countries, royalty, journalists, television presenters, athletes, entertainers, and others’.[44] It includes such tags as ‘cum slut, rape, forced, gangbang, degradation, under skirt, fat, bottle, body writing …’.[45] Under existing Victorian legislation these would be criminalised only because they were published or distributed, but if they were not shared then nothing could be done about it. Introducing similar legislation akin to preventing GenAI CSAM programs is something that could work towards minimising the damage that deepfakes have and that when the public becomes outraged over the proliferance of deepfakes the government can do something.

Revenge Porn and Deepfakes

Revenge Porn is the ‘non-consensual sharing of sexual material’, that is sharing an image, video or audio of a person in a sexual manner using a carriage service (like social media, text, internet, etc) without the person’s consent.[46] It is criminalised under Commonwealth law, as it is an offence to use a carriage service to transmit sexual material without consent.[47] Interestingly, it does have a carveout for AI use and deepfakes, as it includes images, videos or audio edited or entirely created using digital technology to show a realistic but false depiction.[48]

In the age of deepfakes, revenge porn will become a major issue within contemporary Australian society. As previously mentioned, a rudimentary form of deepfake involves superimposing someone’s face onto an already pornographic image, which is known as ‘sexualised photoshopping’.[49] These are happening within schools Australia wide, and they are often perpetrated by children and adults within schools. One popular method is to use a nudification tool that allows for clothes to be digitally removed from images,[50] with young people utilising AI in this method to ‘create and share both real and manipulated explicit content including non-consensual nude images’.[51]

This has become an issue as one in 25 young adults under 18 has either been victimised or has a friend who is the victim of this revenge porn AI trend.[52] Students are creating these materials for all kinds of reasons, including to harass or embarrass their classmates.[53]

Conclusion

While existing legislation is able to address the rise of GenAI in creating CSAM, sexualised deepfakes of people and the increased capability of revenge porn; the fact that it is becoming more prevalent may mean further criminalisation is warranted, akin to manslaughter laws being created to make prosecution easier of king hit attacks. As AI becomes more and more prevalent within society, its potential for harm will not go away. As we see more and more high-profile instances of GenAI porn and CSAM it is likely that the public will demand action from our government and Parliamentarians, a failed bill is evident of this mounting pressure, and it is unlikely to go away anytime soon.

Soon we will have the issue of not knowing whether that girl in the video is real, and whether Will Smith truly ate that spaghetti. The genie is out of the bottle, and no image will ever be truly safe again; when a single click will have you naked, gangbanged and covered in cum, no matter who you are, new legislation needs to be enacted to try and prevent people further misusing this technology.

---

[1] 1958 (Vic) (‘Crimes Act’).

[2] 2004 (Cth) Sch 1.

[3] Crimes Act (n 1) s 51c.

[4] Ibid s 51d.

[5] Ibid s 51g.

[6] Criminal Code Act 1995 (Cth) s 273.6 (‘Criminal Code’).

[7] ‘eSafetyCommissioner’, Generative AI and child safety: A convergence of innovation and exploitation (Web Page, June 2025) (‘eSafetyCommissioner’) citing National Centre for Missing & Exploited Children, CyberTipline Report (Report, 2024) 11.

[8] eSafetyCommissioner (n 7).

[9] Jake Evans, ‘Bill to criminalise AI child abuse apps to be introduced to parliament’, ABC News (28 Jul 2025)<https://www.abc.net.au/news/2025-07-28/kate-chaney-bill-aims-to-criminalise-ai-child-abuse-tools/105571734> (‘Evans’).

[10] Commonwealth, Parliamentary Debates, House of Representatives, 28 July 2025, 284 (Kate Chaney).

[11] ‘Adam Simic’, Digital Abuse and the Challenges of AI-Generated Child Sexual Abuse Material (Web Page 8 October 2025)
<https://koffels.com.au/digital-abuse-and-the-challenges-of-ai-generated-child-sexual-abuse-material/> (‘Simic’).

[12] ‘Australian Federal Police’, Western Sydney man charged for allegedly possessing and producing AI-child abuse material (Web Page, 24 June 2026) 
<https://www.afp.gov.au/news-centre/media-release/western-sydney-man-charged-allegedly-possessing-and-producing-ai-child>.

[13] ‘Australian Federal Police’, Sydney man charged over alleged child-like sex doll importation and producing AI-generated child abuse material (Web Page, 18 October 2025)<https://www.afp.gov.au/news-centre/media-release/sydney-man-charged-over-alleged-child-sex-doll-importation-and-producing>.

[14] ‘Australian Federal Police’, NSW man charged with allegedly producing AI-generated child abuse material and possessing bestiality content (Web Page, 14 August 2025)<https://www.afp.gov.au/news-centre/media-release/nsw-man-charged-allegedly-producing-ai-generated-child-abuse-material-and>.

[15] Ibid.

[16] ‘Australian Federal Police’, Operation Cumberland – Australian duo charged as part of global investigation into AI-generated child abuse material (Web Page, 01 March 2025)<https://www.afp.gov.au/news-centre/media-release/operation-cumberland-australian-duo-charged-part-global-investigation-ai>.

[17] ‘Australian Federal Police’, Victorian man jailed for producing almost 800 AI-generated child abuse images (Web Page, 31 July 2024)
<https://www.afp.gov.au/news-centre/media-release/victorian-man-jailed-producing-almost-800-ai-generated-child-abuse-images>.

[18] ‘Australian Federal Police’, Tasmanian jailed for possessing AI-generated child abuse material (Web Page, 30 March 2024)<https://www.afp.gov.au/news-centre/media-release/tasmanian-jailed-possessing-ai-generated-child-abuse-material>.

[19] Simic (n 11).

[20] Explanatory Memorandum, Criminal Code Amendment (Using Technology to Generate Child Abuse Material) Bill 2025 (Cth) ‘Criminal Code Amendment (Using Technology to Generate Child Abuse Material) Bill’).

[21] Evans (n 9); Criminal Code Amendment (Using Technology to Generate Child Abuse Material) Bill (n 20).

[22] Explanatory Memorandum, Criminal Code Amendment (Deepfake Sexual Material) Bill 2024 (Cth) 3 (‘Criminal Code Amendment (Deepfake Sexual Material) Bill’).

[23] Asher Flynn, Jonathan Clough, and Talani Cooke, ‘Disrupting and Preventing Deepfake Abuse: Exploring Criminal Law Responses to AI-Facilitated Abuse’ in Anastasia Powell, Asher Flynn and Lisa Sugiura (eds), The Palgrave Handbook of Gendered Violence and Technology (Palgrave Macmillan, 2021) 583, 584.

[24] Noelle Martin, ‘Image-Based Sexual Abuse and Deepfakes: A Survivor Turned Activist’s Perspective’ in Anastasia Powell, Asher Flynn and Lisa Sugiura (eds), The Palgrave Handbook of Gendered Violence and Technology (Palgrave Macmillan, 2021) 55, 57 (‘Disrupting and Preventing Deepfake Abuse: Exploring Criminal Law Responses to AI-Facilitated Abuse’).

[25] Ibid.

[26] Ibid 61.

[27] Disrupting and Preventing Deepfake Abuse: Exploring Criminal Law Responses to AI-Facilitated Abuse (n 24) 584.

[28] Qoria, Addressing the Risks of AI-enabled CSAM and Explicit Content in Education (Report, November 2024) 1, 21 (‘Qoria’).

[29] Disrupting and Preventing Deepfake Abuse: Exploring Criminal Law Responses to AI-Facilitated Abuse (n 24) 590.

[30] Criminal Code (n 6) s 474.17.

[31] Ibid s 474.17A.

[32] Crimes Act (n 1) s 21A

[33] R v Wilson [2012].

[34] Disrupting and Preventing Deepfake Abuse: Exploring Criminal Law Responses to AI-Facilitated Abuse (n 24) 591.

[35] Ibid 592.

[36] Summary Offences Act 1966 (Vic) s 41DA (‘Summary’).

[37] Ibid s 41DB.

[38] Ibid s 40.

[39] Disrupting and Preventing Deepfake Abuse: Exploring Criminal Law Responses to AI-Facilitated Abuse (n 24) 593

[40] Classification (Publications, Films and Computer Games) Enforcement Act 1995 (Vic) s 57(1).

[41] Disrupting and Preventing Deepfake Abuse: Exploring Criminal Law Responses to AI-Facilitated Abuse (n 24) 595.

[42] Ibid.

[43] Ibid 598.

[44] ‘Office of Public Affairs’, United States Seizes Domain Names Publishing Nude Digital Forgeries of Famous Women (Web Page, June 12 2026)
<https://www.justice.gov/opa/pr/united-states-seizes-domain-names-publishing-nude-digital-forgeries-famous-women>.

[45] Culture Reframed, Artificial Intelligence, Virtual Reality, and Pornography: How Misogyny Constrains The Future (Report 2024) 1, 11.

[46] Criminal Code Amendment (Deepfake Sexual Material) Bill (n 22) 4.

[47] Criminal Code (n 6) s 474.17A.

[48] Ibid.

[49] Federica Celli, ‘Deepfakes Are Coming: Does Australia Come Prepared?’ (2020) 17(2) Canberra Law Review 193, 197 citing Clare McGlynn, Erika Rackley, and Ruth Houghton, ‘Beyond “Revenge Porn”: The Continuum of Image-Based Sexual Abuse’ (2017) 25(1) Feminist Legal Studies 25, 33.

[50] Qoria (n 28) 16.

[51] Ibid 17.

[52] Cam Wilson, ‘One in 25 teens affected by AI-assisted online sexual abuse, study finds’, ABC News (7 Jul 2026)
<https://www.abc.net.au/news/2026-07-07/1-in-25-children-targeted-with-ai-sexual-imagery/106878000>.

[53] ‘Australian Federal Police’, AFP warns parents over rise in AI-generated child abuse material (Web Page, 28 January 2025)<https://www.afp.gov.au/news-centre/media-release/afp-warns-parents-over-rise-ai-generated-child-abuse-material>.