Sage Journals: Discover world-class research

Abstract

Child sexual abuse material (CSAM) production poses a grave and evolving threat, causing revictimization through the circulation of material for years. The availability of online technology has enabled sophisticated methods of CSAM production and police evasion. The aims of this study were to explore the sociodemographic features of those with a history of CSAM production, explore criminal sub-types of those who produce CSAM, and explore the sociodemographic and criminal features of possible sub-types. Men who came to police attention for CSAM production offenses between 2004–2019 in Victoria, Australia (n = 741) were included. A hierarchical cluster analysis revealed two distinct criminal profiles: (1) a group with specialist offense histories, which were sexually focused (n = 455) and (2) a group with a generalist (n = 286) offending history, including sexual and non-sexual offenses. The frequency of CSAM production offenses across the sample period almost doubled with an increasing number of individuals with a generalist criminal history coming to the attention of police, whilst the number of those with a specialist criminal history remained largely stable. Uncovering these two distinct criminal profiles is a pivotal step toward understanding the nuanced characteristics of individuals responsible for producing CSAM.

Keywords

CSAM child sexual abuse material sexual offending indecent images of children CSAM production

Child sexual abuse material (CSAM) is defined as material depicting the sexual abuse or torture of a child (or representation of a child in a sexually offensive context), whether through text, audio, or image (still and/or moving), Criminal Code Act 1995 (Cth). The production of CSAM has become an escalating phenomenon despite ongoing law reforms and eradication efforts (Seto et al., 2018). Over the past two decades, reports of CSAM to international authorities have increased by an average of 50% annually (Bursztein et al., 2019; Europol, 2020). The production, access, and dissemination of CSAM has increased due to accessibility of the Internet and associated technologies (Interpol, 2020; WeProtect Global Alliance, 2021). Evolving technology provides an efficient infrastructure for uploading, sharing, storing, and concealment; while fostering widespread anonymous contact among people who offend (Coen & Napier, 2022; Owens et al., 2023). Methods of distributing CSAM, such as peer-to-peer networks, newsgroups, and chatrooms (Mitchell et al., 2011; Wolak et al., 2011) can be accessed via mobile platforms, further expanding geographic reach (Bursztein et al., 2019; Internet Watch Foundation, 2021).

Advancements in internet anonymization, encryption, and hard drive eraser software have obscured CSAM distribution from law enforcement (Australian Criminal Intelligence Commission, 2018); necessitating innovative and robust methods of policing, policy, legislation and research, such as undercover operations into CSAM distribution networks (Bleakley, 2019; Thorn, 2014; U.S. Department of Justice, 2016). Identifying child victims depicted in CSAM remains a significant challenge for law enforcement, with most victims remaining unidentified (ECPAT International & Interpol, 2018). Moreover, CSAM production has gravitated towards younger child victims subjected to escalating abuses (torture, sadism and/or bestiality) underscoring the critical importance of ongoing research to understand and combat this expanding crime (Internet Watch Foundation, 2021; Seto et al., 2018).

The profound and enduring repercussions of childhood sexual abuse (CSA) on its survivors has been demonstrated (Cutajar et al., 2010). Survivors of CSAM production have conveyed that the shame, humiliation, and despair stemming from their exploitation not only persisted but intensified over their lives due to ongoing circulation of the material (Martin, 2015; Salter et al., 2021). This ongoing harm underscores the need for comprehensive understanding and intervention to break the chain of exploitation and support survivors in their journey toward healing and recovery. The harm stemming from the use of artificial intelligence (AI) to generate CSAM can be thought of as a spectrum, as it has been reported to cause serious exacerbation of harm to survivors of CSAM through mass production of a child’s image (“famous” children; Internet Watch Foundation, 2024), to “victimless” CSAM - in the sense that the person in the image does not exist. Even though, there may be no direct victim from the creation of an image, it can be argued that harm is perpetuated broadly by increasing the demand for CSAM. However, as Steel (2024) points out, the impact of new methods of CSAM on victimization is underexplored.

The Changing Nature of CSAM Production

Historically, CSAM production has involved individuals actively involving themselves in the physical contact abuse of children during production (Sheehan & Sullivan, 2010); this is no longer the case. Virtual and morphed CSAM does not require physical access to a child (Owens et al., 2023). Virtual CSAM refers to computer-generated content portraying the sexual abuse or exploitation of digitally created children (Levy, 2002). The advent of deepfake technology (replacing a personal likeness with another) has given way to morphed CSAM. This technique involves superimposing a child’s likeness onto existing CSAM, resulting in a fabricated child victim during both the production and distribution phases, and can be easily automated through the widespread adoption of AI software (Nguyen et al., 2022). This emergence of AI-generated CSAM only intensifies the challenge of victim identification for law enforcement.

Increasingly, individuals can also access, direct, and record the live streaming of a child being abused (Napier et al., 2021). Increasing internet speeds across jurisdictional boundaries have been thought to aid the practice of live streaming (Interpol, 2020; WeProtect Global Alliance, 2021). The same is true for self-generated intimate images, commonly referred to as “sexting.” While legislative amendments have resolved issues related to CSAM production charges against consenting minors (Crofts & Lee, 2013; Lee et al., 2013), studies suggest that a significant portion of self-generated intimate images of young people are harvested and distributed within CSAM networks, often by way of grooming and extortion (Internet Watch Foundation, 2012, 2015; Patchin & Hinduja, 2020).

In this rapidly evolving technological landscape, CSAM production is presenting complex challenges, including the concerning ability to produce such material without direct contact offending. The socio-demographic profile of those who engage in CSAM production and other CSAM offenses may well be a product of the population of individuals who had access to the technologies required to produce, distribute and access CSAM (Henshaw et al., 2018; Tomak et al., 2009). The democratization of digital technology means that low and middle-income peoples have and continue to adopt new technology as it emerges (Chen et al., 2022). These shifts challenge our understanding of who engages in CSAM production offenses, requiring us to expand our knowledge of the characteristics and demographics of a potentially broadening spectrum of offending individuals.

Criminal Diversity and Sociodemographic Features of Individuals Who Engage in CSAM Production

Most published literature focuses on individuals who access or distribute CSAM. The few studies investigating the characteristics of individuals charged with CSAM production suggest a sociodemographic profile that is Caucasian, male, mid-30s, in gainful employment and of relatively middle socioeconomic standing (Cale et al., 2021; Salter et al., 2021; Sheehan & Sullivan, 2010; Wolak et al., 2011). This is in stark contrast to the broader population of individuals who engage in CSA or non-sexual criminal behavior, a high proportion of whom are of diverse ethnicity and relatively lower socioeconomic and educational standing (Aslan & Edelmann, 2014; Merdian et al., 2009; Stiernstromer et al., 2022). Across the studies the profile of the Caucasian, middle class, well-educated, 30-year-old male endures. This profile is also found in samples of individuals who access and distribute CSAM but who do not necessarily engage in its production or CSA (Babchishin et al., 2010; Brown & Bricknell, 2018; Merdian et al., 2009; Seto et al., 2010; Wakeling et al., 2011). While a pattern of socioeconomic characteristics is apparent in the limited literature, the criminal diversity of those who have a history of CSAM production remains less clear.

Earlier research indicated that a minority of individuals arrested for CSAM production offending have a history of violence (18%), problems with drugs or alcohol (20%), and/or commit non-sexual crimes (26%; Wolak et al., 2011). This contrasts to most people involved in criminal activities, who tend to exhibit generalized offending patterns (DeLisi & Piquero, 2011; Gottfredson, 2006). Work on those with a history of various types of CSAM offending (Babchishin et al., 2015; Henshaw et al., 2018) has indicated that antisociality is a risk factor for more broader sexual and non-sexual offending. Antisociality encompasses attitudes and traits indicating a disregard for societal norms and standards; the more antisocial, the higher the rates of criminal offending across various types of offenses (i.e., ‘criminal versatility’; Hanson & Bussiere, 1998; Quinsey, 1986). A further notion that may point to a pattern of offending for those with a history of CSAM production is that of offending specialization, when the individual only engages in similar types of crime. Indeed, specialization has been found in a sample of men convicted of accessing CSAM (Howard et al., 2014) and was evident in a sample of men sentenced for sexual offenses, where 51.2% had no convictions for non-sexual crimes (Wortley & Smallbone, 2014). Whether a structure or spectrum (i.e., groups or smooth variations) can be used to describe the offending history of those involved in CSAM production remains unclear, particularly given the changing nature of production offending and facilitating technologies

There are few published studies of CSAM production in the literature. Wolak and Colleagues (2005) studied 122 individuals who were arrested for CSAM production in the US from July 2000 to June 2021, which was later extended upon by Wolak et al. (2011) who compared the 2000/2001 sample to an additional sample (n = 197) from 2006. Both studies screened out a large proportion of cases in which the victim was not identified, thereby restricting findings to a very specific production cohort. Wolak and Colleagues (2011) reported a decrease in full time employment and those with college or technical training between the groups. There were also changes in offending histories between the groups; with increases in arrests for non-sexual crime and being a registered sex offender. These findings point to lower socioeconomic groups engaging in CSAM production, with more diverse offending behaviors in general over time. However, as this work is quite dated, it remains unknown whether these findings would apply to more recent samples involved in CSAM production.

The Current Study

This study builds on the work of Wolak and Colleagues (2005; 2011) by including the entire population of 741 men in Victoria, Australia, who came to the attention of Victoria Police for CSAM production over a 16-year period (2004–2019). Australian data and local state and Commonwealth legislation was used. The current study aims to: (1) describe the demographic backgrounds and lifetime offending histories of men charged with CSAM production offenses, (2) explore the possibility of CSAM production sub-types based on lifetime criminal history, and (3) describe the associated sociodemographic and lifetime criminal features of any sub-types identified. The aims of this paper were developed once researchers had screened law enforcement data and concluded that the production of CSAM required exploration as a separate topic from other online child sexual exploitation offending. To date the literature indicates that those who engage in CSAM production have been treated as a homogenous group. However, it is unknown whether this profile has remained consistent since the work of Wolak and Colleagues (2005; 2011). To the knowledge of the authors, any latent criminal subtypes of individuals who have produced CSAM have not been explored quantitatively or compared on sociodemographic or criminal grounds.

Method

Design

The current study is part of a broader program of research focused on online child sexual exploitation funded by an Australian Research Council Linkage Project grant (LP180100090) in partnership with Victoria Police, Corrections Victoria, the Australian Institute of Criminology, and Monash University. Data examined in this research were obtained from secondary pre-existing electronic databases that contain criminal records (charges and convictions), the Law Enforcement Assistance Program (LEAP) maintained by Victoria Police, and the National Police Reference System (NPRS) maintained by the Australian Criminal Intelligence Commission. No active participants were involved in the research. The authors take responsibility for the integrity of the data, the accuracy of the data analyses, and have made every effort to avoid inflating statistically significant results. Ethical approval for this research was provided by the Justice Human Research Ethics Committee (CF/20/17580) and Swinbunre Human Research Ethics Committee (20215770–6587).

Sample

The study sample comprised the population of adult males who had received a formal charge or been reported to Victoria Police for suspected CSAM production in Victoria between 2004–2019 (inclusive) based on their LEAP history (n = 741). Individuals may or may not have proceed to court or been convicted for the CSAM production offenses that brought them to the attention of police (this information was not included in the data). Study inclusion required people to have had at least one record for CSAM production in their LEAP history (i.e., lifetime offending) with a processing date between 2004 and 2019. For inclusion in this study, CSAM production was defined as offences involving digitally facilitated creation of CSAM (actioned or attempted). Given the broad definition within Australian legislation, creation of material may occur online or in-person, and may or may not involve real children, see the Appendix for a list of charge descriptions and relevant Australian legislation. Specific details about cases were not available to researchers, such as how images were generated.

Individuals who were suspected of CSAM production when they were minors (under the age of 18 years) but not as adults were excluded as they are treated differently to adults under Victoria and Australian criminal law (Crimes Amendment (Sexual Offenses and Other Matters) Act 2014 (Vic); Richards, 2011). Individuals who did not have any record of CSAM production were excluded and women were excluded. Women were also excluded because patterns and drivers of sexual offending for women tend to differ from men (e.g. Freeman & Sandler, 2008), which would have confounded the results. In total, 6670 individuals did not have an online CSAM production offending history and were excluded from the original 7411 received by Swinburne researchers.

Data Sources and Linkage Procedure

LEAP and NPRS are dynamic policing databases that enable the documentation of criminal activity locally in Victoria and interstate, respectively (Australian Criminal Intelligence Commission, 2023; Victoria Police, 2022). Individuals were first identified within the LEAP database and then the data were linked to the NPRS to produce a dataset of each individual’s national criminal history up until 2022. Table 1 shows the variables extracted from the LEAP and NPRS databases and how they were used in this study.

Table 1.

Source of Variables, Their Description and Use.

Database	Variable	Description	Usage and details
LEAP and NPRS	Study ID	Numerical deidentified participant study identifier	Used to link individuals across data bases
LEAP	Date of Birth	Month and year of birth	Used to calculate age at which offense was committed
LEAP	Post code	Last known residential postcode of the individual	Used to create last known socioeconomic and remoteness index by mapping post codes on to standardised ABS categories
LEAP	Charge code and charge description	Criminal charge listed within an incident of police activity	Used to identify specific child sexual offending categories of interest in this study. Codes were also mapped onto the ANZSOC from the ABS. The LEAP data was used to capture lifetime offending history in Victoria
LEAP and NPRS	Processing date	The administrative date at which a case is finalised by police or completed by data base administrators	Used to calculate age at offending. What determines the processing date between sources differs due to administrative processes and is a limitation to the linkage procedure
NPRS	ANZSOC codes and child sexual offending categories of interest	The 16 ANSOC codes, plus additional sub-categories of child sexual offenses of interest to this study	Used to capture lifetime offending history from jurisdictions other than Victoria. This data included offending histories from other Australian states and the Australian Federal Police

Offense rates from policing charge data are less conservative (i.e., more inclusive) than convictions due to the presence of unproven crimes in the criminal justice system (Lievore, 2004; Payne, 2007). However, not all offenses come to police attention, this is especially true for sexual offenses, where under 19% of sexual assaults are reported and only 1.3% reach court (Gelb, 2007). Thus, court records likely underestimate true offense rates, leaving police data as the most viable approximation available.

To provide context for descriptives, researchers accessed the Australian Bureau of statistics (ABS) state population data which is available to the public on their website (ABS, 2023a, 2023b). Victorian Population demographic summary statistics for Socio-Economic Quintile Indexes for Areas (SEIFA) 2016 and Australian Statistical Geography Standard – Remoteness Area (ASGS-RA) were used.

Demographic and Offense Classification

Demographics

Socio-Economic Quintile Indexes for Areas (SEIFA) 2016

The SEIFA (ABS, 2021a) is a five-level measure of social and economic indicators of disadvantage by geographical region in Australia (1 = most disadvantaged, 5 = least disadvantaged). The SEIFA offers a consistent estimate of socioeconomic index by postcode and thus has become a standard in much Australian public health and criminology research, due to its utility in comparing between sample populations (Allard et al., 2012; Karriker-Jaffe, 2011; Walker & Hiller, 2007). The Australian Bureau of Statistics publishes up to date estimates of the Victorian population by residential Socio-economic Quintile Index by year (ABS, 2023b).

Australian Statistical Geography Standard – Remoteness Area (ASGS-RA)

The ASGS-RA (ABS, 2021b) is a four-level measure of remoteness distribution and relative access to services for every geographic area in Australia based on an overall rank score based on postcode (RA1 = Major Cities of Australia, RA4 = Very Remote Australia). The Australian Bureau of Statistics publishes up to date estimates of the Victorian population by residential remoteness code by year (ABS, 2023a). The ASGS-RA has been used alongside SEIFA and the Australian and New Zealand Standard Offense Classification (ANZSOC) in geographical criminology research to assess crime concentration (Allard et al., 2012).

Offense Classification

Australian and New Zealand Standard Offense Classification (ANZSOC) 2011 (3rd ed)

The ANZSOC is a three-dimension criminal behavior classification standard for the analysis of crime and justice statistics (ABS, 2011). Each level provides greater classification fidelity of the target behavior: Divisions (03 = sexual assault and related offenses), Subdivisions (320 = non-assaultive sexual offenses) and Groups (0322 = CSAM offenses). The ANZSOC is used by Australian and New Zealand police and criminal courts, providing a standardized framework for classifying criminal offenses. A total of 751 unique charges were listed within the data provided by Victoria Police from LEAP and the Australian Criminal Intelligence Commission from the NPRS. All charges were integrated into the broader offense categories, as defined by the ANZSOC.

Sexual Offense Classification

The ANZSOC does not have the specificity required to operationalize sexual crimes for the purposes of this research. For example, all CSAM charges are subsumed under ANZSOC code 0322 whether that be CSAM possession, distribution, or production. Furthermore, all penetrative sexual assault charges are subsumed under ANZSOC code 0311 whether perpetrated against an adult or child. Therefore, all sexual assault and related offenses were manually reclassified by the research team according to their unique charge description. Table 2 defines the five categories of sexual offenses examined within this research.

Table 2.

Sexual Offense Categorisation Scheme.

Sexual offense category		Definition	Charge examples
1	Contact child sexual assault	Offenses that involve the sexual physical touching of a child, or oneself or another in the presence of a child	Sexual penetration with a child, indecent act with a child, persistent sexual abuse of a child, gross indecency with a child
2	Grooming and sexual chat	Offenses that are intended to coerce or encourage a child into engaging in sexual behavior	Use of a carriage service to procure, groom, or solicit a child for a sexual act, use an on-line service to transmit sexual material to a child
3	Possess/access CSAM	Offenses whereby an individual knowingly accesses or possesses CSAM	Knowingly possess CSAM, access CSAM using a carriage service, cause CSAM to be transmitted to self, aggravated overseas CSAM offenses
5	Child sexual servitude	Offenses that involve causing a child to be sexually abused for financial gain	Cause or induce a child to act as a sex worker, enter into an agreement for a child sex service
6	CSAM production	Offenses involving the creation of material (visual, audio or text-based) that depict or describe the sexual abuse of a child	Make/produce CSAM, involve a child in the production of CSAM, attempt to produce CSAM

Note. Attempted charges have been included as if they happened, as Victoria Police have clarified that without police intervention the offense would have likely been executed unimpeded.

Data Screening

The data were largely complete across the demographic variables. However, a small number of last known postcodes (i.e., zip codes) were missing (3.37%), which were required for matching socioeconomic index and remoteness area. Investigation of the extracted variables provided no evidence to indicate the missing values were Missing Not at Random. The missingness could not be accounted for by any other variable in the dataset which would indicate that it was Missing Completely at Random rather than Missing at Random. As the missingness of the postcodes was <5% of all cases, a missing value was created and excluded from the relevant Chi-square tests. This technique was applied given it was the most straightforward process, as other techniques (e.g., mean substitution, imputation, expectation maximization, etc.) are equally likely to provide similar results with this level of missingness (Tabachnick & Fidell, 2013).

Data Analysis

To address aim one, descriptive statistics for sociodemographic and lifetime offending characteristics were conducted on the population of males who came to the attention of Victoria Police for CSAM production offenses between 2004–2019 (inclusive). Demographic variables included age at first criminal charge, age at first sexual charge, age at first CSAM production charge, last known socio-economic index, and last known remoteness category. Offending variables included the 16 ANZSOC codes (i.e., offense types), contact child sexual offenses, grooming and sexual chat, possession and accessing CSAM, child sexual servitude and CSAM production.

To provide context and meaning to the descriptive statistics, we compared the sample who have a record of CSAM production to the Victorian population. Chi square goodness of fit tests were conducted to explore whether the CSAM production sample reflects the socioeconomic index levels and remoteness codes of the general population.

To explore the trend in CSAM production over time, (i.e., did CSAM production offending increase or decrease over time in Victoria), a linear regression was performed. Year was entered into the analysis as a continuous independent variable and annual number of CSAM production charges was entered as the dependent variable.

Aim two, to explore the possibility of CSAM production sub-types based on lifetime criminal history, was addressed with a hierarchical cluster analysis using Ward’s (1963) method with the Chi-square distance for count data. To examine whether the diversity and frequency of criminal offending could provide an indication of how many clusters are required to describe the sample, the count of charges for each of the 16 ANZSOC categories from the national offending history of each individual was entered into the analysis. The analysis would reveal any cluster groups based on similar offending histories (Everitt et al., 2001; Kaufman & Rousseeuw, 1990). Ward’s (1963) method employing the Chi-square for count data minimizes the within-cluster variance of cases when merging clusters, which is particularly useful when the variance is expressed as a measure of the count dissimilarity across ANZSOC divisions (Jain & Dubes, 1988; Ward, 1963). The hierarchical cluster analysis retains its robustness as an exploratory statistical technique providing that all included variables are relevant to offending characteristics of CSAM production.

As subtypes were identified in the analyses undertaken for Aim two, Aim three was addressed by describing the associated sociodemographic and criminal features of identified sub-types. Individuals were assigned to cluster membership (specialist and generalist) for further exploration of sociodemographic and offending history features. Chi square tests of independence were conducted to compare the frequency of sociodemographic characteristics between the two clusters.

To address the question of whether the clusters were useful in predicting the frequency of different types of offending, a series of 21 Univariate Negative Binomial Regressions were conducted. The use of a negative binomial distribution was required due to the large number of zero values in the count data. Offending type was entered as the dependent variable and cluster as the predictor variable for all 21 regressions, with a Bonferroni correction used (alpha = .002) as the threshold of statistical significance.

To assess whether the change over time in the number of annual CSAM production charges differ between clusters a generalized linear model with a Poisson distribution, was run. Annual CSAM Production count was entered as the dependent variable, with Year as a covariate and a Year by Cluster interaction as a predictor variable.

Results

Description of Sociodemographic Features and Offending History

Aim one was to describe those who produce CSAM. Included in the study were 741 individuals that came to the attention of Victoria Police for CSAM production within 2004–2019. Most were aged 18–25 years old when their first charge of CSAM production, first criminal charge and first sexual charge were recorded (see Table 3). Based on the last known and valid postcodes (n = 716, 96.63%), individuals tended to live in the lowest (n = 180, 24.29%) and middle upper (n = 159, 21.46%) socio-economic SEIFA indexes and resided in major cities (n = 435, 58.75%). The fewest individuals tended to be in the highest socioeconomic SEIFA index (n = 118, 15.92%) and live in outer regional and very remote areas (n = 52, 7.02%).

Table 3.

Descriptive Statistics for Individuals Known to Victoria Police for CSAM Production Between 2004 and 2019.

Demographic variable	N	%
Age at first criminal charge
<18	91	12.28
18–25	176	23.75
26–32	113	15.25
33–40	105	14.17
41–48	94	12.69
49–56	91	12.28
57+	71	9.58
Age at first sexual charge
<18	17	2.29
18–25	174	23.48
26–32	125	16.87
33–40	130	17.54
41–48	113	15.25
49–56	103	13.90
57+	79	10.66
Age at first CSAM Production charge
18–25	173	23.35
26–32	109	14.71
33–40	138	18.62
41–48	119	16.06
49–56	111	14.98
57+	91	12.28

Chi-square Goodness of Fit Tests revealed significant differences between those with a history of CSAM production offending and the general male Victorian population for both socio-economic index (χ² (4) = 17.89, p < .001) and remoteness code (χ² (2) = 307.39, p < .001). Those with a history of CSAM production offending tended to reside in the lowest and middle upper socio-economic indexed which was significantly different to the general Victorian population who tended towards the middle and highest indexes (see Figure 1). While those sampled tended to reside in major cities as is the case for most Victorians, a higher proportion tended to reside in inner regional Victoria than the general Victorian population (see Figure 2).

Figure 1.

Last known socio-economic index of sample vs census population for 2021.

Figure 2.

Last known remoteness code of sample vs census population for 2021.

Victoria Police recorded a total of 770 contacts with the sample for CSAM production offending between 2004–2019 (see Figure 3). The influence of time (years) on CSAM production charges (annual CSAM production charge count) was significant (F (1,14) = 10.27, p = .006). Forty-two percent of the variance in annual number of CSAM production charges can be explained by years. For every additional year the number of CSAM production charges increase by 1.05 (t = 3.21, p = .006).

Figure 3.

Frequency of CSAM production police activity/charges processed by Victoria Police 2004–2019.

A total of 606 (81.78%) individuals had engaged in non-sexual offenses across criminal histories as recorded in LEAP and NPRS (1961–2022; see Table 4). More than half had offenses against government procedures, security, and operations, and approximately a third had offenses categorised as act intended to cause injury.

Table 4.

Descriptive Statistics of the National Offending History of Those Known to Police in Victoria for CSAM Production Between 2004–2019.

Offense type	N	%	Median	Range
ANZSOC offense category
01 Homicide and related offenses	5	0.68	0.00	0–2
02 Acts intended to cause injury	272	36.71	0.00	0–15
03 Sexual assault and related offenses	471	100.00	2.00	1–10
04 Dangerous or negligent acts endangering persons	59	7.96	0.00	0–6
05 Abduction, harassment and other offenses against the person	147	23.48	0.00	0–18
06 Robbery, extortion and related offenses	27	3.64	0.00	0–3
07 Unlawful entry with intent/burglary, break and enter	87	11.74	0.00	0–7
08 Theft and related offenses	211	28.48	0.00	0–27
09 Fraud, deception and related offenses	92	12.42	0.00	0–1
10 Illicit drug offenses	157	21.20	0.00	0–14
11 Weapons and explosives offenses	149	20.11	0.00	0–16
12 Property damage and environmental pollution	143	19.30	0.00	0–24
13 Public order offenses	189	25.51	0.00	0–16
14 Traffic and vehicle regulatory offenses	71	9.58	0.00	0–15
15 Offenses against government procedures, security and operations	446	60.19	1.00	0–30
16 Miscellaneous offenses	105	14.17	0.00	0–5
Sexual offenses
Contact child sexual offenses	344	46.42	0.00	0–15
Grooming and sexual chat	77	10.39	0.00	0–11
Possess/access CSAM	671	90.55	1.00	0–9
Child sexual servitude	9	1.21	0.00	0–2
CSAM production	741	100.00	1.00	1–3

Note. n = 741; the table includes LEAP data – offending recorded by Victoria Police and NPRS data – offending compiled from other Australian states and the APF. Median (a measure of central tendency) and Range (the lowest and highest values) were reported to describe the distribution of offences and highlight their extreme positive skew caused by the inflated number of zeros in the count data.

Most individuals had at least one charge of accessing or possessing CSAM, and just under half had a charge of contact child sexual offenses. A relatively small number of individuals had a record of grooming and sexual chat, or sexual servitude offenses against children. Only nine individuals (1.21%) had CSAM production as their only criminal offense.

Hierarchical Cluster Analysis Based on Charge Count for Each ANZSOC Division

Aim two was to explore sub-types of those who engage in CSAM production. Inspection of the hierarchical cluster analysis dendrogram resulted in two final clusters (cut at 25 rescaled distance) that encompassed all cases in the data (see Figure 4). Whilst a three-cluster solution was investigated (cut at 20 rescaled distance) cluster group membership was largely based on offenses against government procedures, security and operations and did not assist in determining CSAM production typology. A two-cluster solution was preferred as it illustrates the greatest rescaled distance between all other cluster solutions, which represents two groups with the least variability. The two-cluster solution indicated that cluster assignment was dependent on whether individuals have specialist criminal histories (n = 455, the first cluster consisting of those who focused on sexual offending) or generalist histories (n = 286, the second cluster consisting of those who engaged in a broad spectrum of offending types). To explore effects of charge seriousness, time incarcerated and charge count on clusters, a latent class analysis was run with the ANZSOC codes as binary variables. Two clusters were apparent, and the subgroups specialist and generalist could be identified when ANZSOC counts were explored by subgroup membership (see the Supplement Analysis In The Appendix).

Figure 4.

Hierarchical Cluster Analysis Dendrogram using Ward’s Method Showing Two Clusters. Note: Cluster A = Specialist criminal history, Cluster B = Generalist criminal history.

Sociodemographic and Offending History: Comparisons Between Sub-types

Aim three was to explore sociodemographic and offending history features based on the results of the cluster analysis. Chi-square tests of independence revealed significant differences between the two groups across several sociodemographic indicators (see Table 5). Generalists were significantly more likely than specialists to have their first criminal charge under the age of 18, but less likely to have their first criminal charge at an age of above 48. For age of first sexual charge, those with specialist criminal histories were more likely to first come to the attention of police after the age of 56 than generalists. For age at first CSAM production charge, individuals with a specialist criminal history were less likely to come to the attention of police for CSAM production during the ages 33–40 years but more likely to come to the attention of police for CSAM production charges above the age of 57 than individuals with a generalist history.

Table 5.

Sociodemographic Characteristics Based on CSAM Production Sub-type Membership.

Demographic variable	Specialist n (%)	Generalist n (%)	df	χ2
Age at first criminal charge (any type)			6	86.00***
<18	19 (4.18)	72 (25.17)^#
18–24	109 (23.96)	67 (23.43)
25–32	71 (15.60)	42 (14.69)
33–40	67 (14.73)	38 (13.29)
41–48	61 (13.41)	33 (11.54)
49–56	69 (15.16)^#	22 (7.69)
57+	59 (12.97)^#	12 (4.20)
Age at first sexual charge			6	18.352**
<18	7 (1.45)	10 (3.50)
18–24	106 (23.30)	68 (23.78)
25–32	77 (16.92)	48 (16.78)
33–40	68 (14.95)	62 (21.68)
41–48	67 (14.73)	46 (16.08)
49–56	68 (14.95)	35 (12.24)
57+	62 (13.63)^#	17 (5.94)
Age at first CSAM Production charge			5	16.387***
18–24	108 (23.74)	65 (22.73)
25–32	65 (14.29)	44 (15.38)
33–40	71 (15.60)^#	67 (23.43)
41–48	69 (15.16)	50 (17.48)
49–56	72 (15.38)	39 (13.64)
57+	70 (15.38)^#	21 (7.34)
Last known socio-economic index			4	8.15*
Lowest index	94 (21.46)	85 (30.58)^#
Lower middle index	82 (18.72)	51 (18.35)
Middle index	81 (18.49)	46 (16.55)
Middle upper index	103 (23.52)	56 (20.14)
Highest Index	78 (17.81)	40 (14.39)
Missing^a	17 (3.74)	8 (2.80)
Last known remoteness category			2	1.50
Major city	273 (62.33)	162 (58.06)
Inner regional	136 (31.05)	94 (33.69)
Outer regional & very remote^a	29 (6.62)	23 (8.24)
Missing^a	17 (3.74)	8 (2.80)

Note. All significance testing was completed subject to the Fisher-Freeman Halton Exact Test. ^# significantly higher row differences indicated by adjusted residuals higher than 1.96, *p < .05. **p < .01. ***p < .001.

^aMissing not included in test.

Testing revealed a significant difference in socioeconomic index based on group membership, those with a generalist criminal history being more likely to reside in the lowest index compared to those with a specialist criminal history. The remoteness location of where an individual resided did not significantly differ according to cluster membership (see Table 5).

A series of univariate negative binomial regressions were conducted with the number of offenses for each ANSOC division as the dependent variable and cluster (specialist = 0, generalist = 1) as the independent variable (see Table 6). Individuals with a generalist criminal history were significantly more likely to come to the attention of police across a range of different non-sexual offense types. Cluster membership did not appear to predict the likelihood of homicide related offenses, likely due to very low numbers of homicides. Cluster membership also did not predict types of child sexual offenses. Those with a generalist criminal history tended to commit a broader range of offenses in greater frequency than individuals whose criminal history was categorized as specialist.

Table 6.

Mean Number of Offenses.

			Specialist cluster = 0		Generalist cluster = 1
Dependant variable offense type	Wald (χ)	p	Mean	95% CI	Mean	95% CI
ANZSOC offense category
01 Homicide and related offenses	2.311	.128	0.002	0.002–0.018	0.013	0.004–0.041
02 Acts intended to cause injury	84.710	<.001	0.378	0.293–0.486	1.755	1.425–2.162
03 Sexual assault and related offenses	0.054	.817	1.936	1.815–2.066	1.913	1.762–2.705
04 Dangerous or negligent acts endangering persons	35.664	<.001	0.006	0.001–0.022	0.280	0.215–0.364
05 Abduction, harassment, and other offenses against the person	18.374	<.001	0.211	0.144–0.310	0.323	0.458–0.875
06 Robbery, extortion, and related offenses	17.395	<.001	0.013	0.005–0.034	0.122	0.080–0.186
07 Unlawful entry with intent/burglary, break and enter	44.358	<.001	0.022	0.009–0.052	0.486	0.367–0.643
08 Theft and related offenses	177.411	<.001	0.119	0.080–0.176	2.325	1.917–2.821
09 Fraud, deception, and related offenses	63.414	<.001	0.020	0.009–0.043	0.545	0.433–0.688
10 Illicit drug offenses	123.217	<.001	0.042	0.023–0.075	1.323	1.106–1.630
11 Weapons and explosives offenses	67.806	<.001	0.134	0.096–0.187	0.727	0.581–0.909
12 Property damage and environmental pollution	66.203	<.001	0.116	0.074–0.183	1.014	0.782–1.314
13 Public order offenses	94.609	<.001	0.180	0.134–0.242	1.073	0.876–1.315
14 Traffic and vehicle regulatory offenses	12.307	<.001	0.097	0.056–0.168	0.332	0.220–0.502
15 Offenses against government procedures, security, and operations	27.760	<.001	1.450	1.428–1.686	2.675	2.254–3.174
16 Miscellaneous offenses	6.471	.011	0.138	0.104–0.184	0.234	0.176–0.312
Sexual offenses
Contact child sexual offenses	0.857	.355	1.178	1.027–1.351	1.059	0.887–1.265
Grooming and sexual chat	2.299	.129	0.261	0.190–0.359	0.167	0.104–0.271
Possess/access CSAM	0.041	.840	1.455	1.351–1.567	1.437	1.308–1.579
Child sexual servitude	0.812	.368	0.013	0.005–0.035	0.024	0.010–0.061
CSAM production	0.582	.446	1.123	1.089–1.158	1.101	1.059–1.145

Note. df = 1 for all tests; n = 741; specialist n = 455 and generalist n = 286, Bonferroni correction (Alpha = .002).

A generalized linear model with a Poisson distribution was run with annual CSAM production count as the dependant variable (M = 24.062, SD = 7.636), year as a covariate (2004–2019), cluster (specialist = 0, generalist = 1) as a main effect and the year × cluster interaction. Cluster significantly contributed to the prediction of annual CSAM production charge count (Wald chi-square = 8.029, df = 1, p = .005) in the model, as did year (Wald chi-square = 15.120, df = 1, p < .001). The interaction Year x Cluster also significantly predicted annual CSAM charge count (Wald Chi-square = 7.948, df = 1, p = .005), indicating a steeper slope for generalist than specialist offenders with generalist offenders starting from a lower base in 2004 (see Figure 5).

Figure 5.

Frequency of CSAM production incidents based on sub-type membership processed by Victoria Police 2004–2019.

Discussion

This study challenges previous research into CSAM production (Wolak et al., 2005, 2011), which was based on interviews of investigators about individuals arrested for CSAM production in the United States of America. They found that most individuals who engage in this behavior are overwhelmingly middle aged, middle-class men, with low rates of other offending across their criminal histories (Cale et al., 2021; Sheehan & Sullivan, 2010). However, a proportion did fit that criminal and sociodemographic profile which is also shared with individuals who engage in CSAM use more broadly (Babchishin et al., 2010; Bourke & Hernandez, 2009; Brown & Bricknell, 2018; Merdian et al., 2009; Seto et al., 2010; Wakeling et al., 2011). In the current sample, a large proportion were characterized by the lowest socioeconomic index, were younger with regards to their first police contact for CSAM production or sexual offense and had a record of diverse and violent criminal offending, including CSA.

Results revealed that the instances of CSAM production as recorded by Victoria Police have significantly increased between 2004–2019 and reflects increasing online child sexual abuse rates internationally (Bursztein et al., 2019; Europol, 2020). Specific peaks and troughs within Figure 3 were not statistically investigated to reduce the chances of Type 1 errors. However, increased police intervention into CSAM production across this period potentially corresponds with two undercover operations into online CSAM networks by Taskforce ARGOS of the Queensland Police: Operation Achillies (2005–2007) and Operation Artimus (2015–2017; Bleakley, 2019). These operations may be a necessary strategy in combatting the increasingly difficult issue of CSAM production and distribution in the modern age.

Contrary to prior research, which conceptualizes those who engage in CSAM production as one group (Cale et al., 2021; Sheehan & Sullivan, 2010; Wolak et al., 2005), our results showed two distinct criminal sub-types: those who are specialists and those who are generalists, with most individuals belonging to the former subtype. Generalists tended to commit the broadest range of violent and non-violent crimes and in greater frequency. No meaningful group differences were observed based on sexual crimes, pointing to the notion that those engaged in CSAM production also tend to engage in other child sexual abuse activities. Wolak and Colleagues (2011) observed a significant increase of individuals with generalist criminal histories engaging in CSAM production between 2000 (26%) and 2006 (43%). We may be observing a natural continuation of this trend. The confirmation of these two groups over time give support to the specialist/generalist categorization within those who engage in CSAM production (DeLisi & Piquero, 2011; Gottfredson, 2006; Howard et al., 2014; Wortley & Smallbone, 2014).

Unsurprisingly, those categorized in the generalist sub-type tended to commence any criminal offending earlier in life and be less inclined to commit their first offense in later life. Further differences were observed through the specialist sub-type tending to be less likely to live in lower socioeconomic areas when compared to the generalist sub-type. These findings challenge the long-held archetype of the middle-aged, middle-class male who engages in CSAM production and/or access (Babchishin et al., 2010; Bourke & Hernandez, 2009; Brown & Bricknell, 2018; Merdian et al., 2009; Seto et al., 2010; Wakeling et al., 2011; Wolak et al., 2011).

As for the age of first CSAM production charge, those with a specialist criminal history were less likely to be charged with their first CSAM production charge in the 33–49 year age group and more likely in later life. Both sub-groups may be more likely to have access to children during the 33–49-year age group than any other, especially their own children.

Our results show an emergent group of individuals from a lower sociodemographic profile, with more frequent and diverse criminal characteristics entering into CSAM production since 2004. The sub-group of generalists show similar characteristics to those who engage in CSA more broadly (Aslan & Edelmann, 2014; Babchishin et al., 2015; Merdian et al., 2009; Stiernströmer et al., 2022). A review of the studies employed by Wolak and colleagues (2005; 2011) may suggest the generalist CSAM producer sub-type had begun engaging in CSAM production between 2000 and 2006 given the changes in characteristics observed between these two periods. It is suggested that the rapid proliferation of digital technology and changes to the methods of CSAM production over the 16 years of sampling have allowed a younger, more diverse group of individuals, driven by high antisociality, who previously had engaged in CSA, the ability to document and distribute their crimes through the production of CSAM. This explanation of our results not only accounts for what was found, but also for what was not found i.e., types of child sexual abuse was not predicted by the specialist and generalist sub-types.

Implications

It is important to recognize that individuals involved in CSAM production should no longer be considered a largely homogenous group of middle-aged, middle-class men who use CSA as the primary means of CSAM production. While further research is required to replicate and extend the findings of this study, our findings point to the possibility that generalists and specialists possess distinct motivations, risk profiles and criminogenic needs. Based on prior research (Babchishin et al., 2015; Henshaw et al., 2018; Seto, 2019) we might speculate that specialists (the minority of people who produce CSAM) may be more motivated by sexual interests and gratification, while generalists (the majority) may be more driven by a broader pattern of criminogenic needs (e.g. antisocial tendencies and sexual interests). Meaning that criminal justice management, intervention and prevention strategies for the generalist subgroup may require more resources and attention than the specailist subgroup. Further, these decisions could be made earlier, possibly when an individual is reported to police for CSAM production, as their offending history will indicate the indivials need for future services.

Identifying subgroups of those who produce CSAM provides an important opportunity to prevent CSAM production offenses – particularly among those driven primarily by antisociality – through standard management and intervention strategies targeting general criminogenic needs. Should future research replicate the sub-types and differentiating factors identified in this study, closer examinations of the offending trajectories and associated predictors of CSAM production sub-types would be of benefit to prevention efforts.

Strengths, Limitations and Future Research

This study represents a necessary update to the growing body of work examining the individuals who engage in CSAM related offenses (Babchishin et al., 2010; Bourke & Hernandez, 2009; Brown & Bricknell, 2018; Henshaw et al., 2018; Merdian et al., 2009; Seto et al., 2010; Wakeling et al., 2011). Our study adds to the research on the characteristics of the men who produce this content as a group worthy of study (Wolak et al., 2005, 2011). To our knowledge, the current study utilizes the largest and most representative sample of individuals detected for CSAM production across studies internationally. However, the study is preliminary, and several limitations and directions for further research must be noted.

The limitations of secondary law enforcement data have been discussed, and due to the fidelity of the dataset, many socio-demographic variables could not be replicated from previous research: such as marital status, drug, and alcohol use, whether the individual lives with a child, or their employment status, education, and ethnicity. The unclear time point at which postcode was captured and effect of data extraction date on age and duration of lifetime offending may have influenced the findings. Furthermore, this study can only characterize individuals who have come to the attention of police and not the larger majority of individuals who produce CSAM undetected. This study is exploratory, and the finding require replication to improve confidence in them.

This study lays the foundation for psychological research on what motivates specialist and generalist sub-types to produce CSAM. The application and usefulness of the specialist and generalist sub-types to mental health and offender management is recommend for future research. For example, confirming differences in age and important psychological constructs between specialist and generalist sub-types may improve accuracy in risk assessment. Further research should consider exploring whether clusters form around sex – men and women. Women who produce CSAM were not included in this study and potential underlying groups in a female cohort should also be explored.

Conclusion

This study provides an important update to the small body of research on individuals who have produced CSAM (Wolak et al., 2005, 2011). The accessibility of digital technology has altered the way we relate to and connect with one another but has created significant challenges in curbing the online sexual exploitation of children. It is essential that this crime be understood with regards to the changing methods of CSAM production and the diversifying group of individuals who commit these crimes. This exploration into unmasking the men that produce CSAM in our modern age has challenged the persistent profile that these crimes are largely committed by middle-aged, middle-income men who are specialists in their criminal activity. Future research into CSAM production should consider the diversity of individual criminal offending given the democratization of digital technology and the reduced barrier to producing digital depictions of CSA.

Supplemental Material

Supplemental Material - Unmasking the Men Who Produce Child Sexual Abuse Material (CSAM): Criminal Diversity and Sociodemographic Characteristics

Supplemental Material for Unmasking the Men Who Produce Child Sexual Abuse Material (CSAM): Criminal Diversity and Sociodemographic Characteristics by Daniel King, Reneta Slikboer, Marie Henshaw, Denny Meyer and James R. P. Ogloff in Sexual Abuse

Footnotes

ORCID iDs

Reneta Slikboer

Marie Henshaw

Denny Meyer

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The current study forms part of an Australian Research Council Linkage Project (LP180100090) in partnership with Victoria Police, Corrections Victoria, the Australian Institute of Criminology, and Monash University. The views expressed herein are solely those of the authors, and do not necessarily reflect the views of our partners.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Legal agreements with data providers and ethics approval prohibit the wider sharing of the data analysed in this study.*

Supplemental Material

Supplemental material for this article is available online.

References

Allard

Chrzanowski

Stewart

(2012). Targeting crime prevention to reduce offending: Identifying communities that generate chronic and costly offenders. In: Trends & Issues in Crime and Criminal Justice, Vol. 445. Australian Institute of Criminology. https://doi.org/10.52922/ti250293

Aslan

Edelmann

(2014). Demographic and offense characteristics: A comparison of sex offenders convicted of possessing indecent images of children, committing contact sex offenses or both offenses. Journal of Forensic Psychiatry and Psychology, 25(2), 121–134. https://doi.org/10.1080/14789949.2014.884618

Australian Bureau of Statistics . (2011). Australian and New Zealand Standard Offence Classification (ANZSOC). https://www.abs.gov.au/statistics/classifications/australian-and-newzealandstandard-offence-classification-anzsoc/latest-release

Australian Bureau of Statistics . (2021a). Socio-Economic Indexes for Areas (SEIFA), Australia. https://www.abs.gov.au/statistics/people/people-andcommunities/socio-economic-indexesareas-seifa-australia/latest-release

Australian Bureau of Statistics . (2021b). Australian Statistical Geography Standard – Remoteness Area (ASGS-RA). Allocation files. (abs.gov.au).

Australian Bureau of Statistics . (2023a). Regional population, 2021-22. https://www.abs.gov.au/statistics/people/population/regional-population/latest-release

Australian Bureau of Statistics . (2023b). Socio-Economic Indexes for Australia (SEIFA), 2021. https://www.abs.gov.au/statistics/people/people-and-communities/socio-economic-indexes-areas-seifa-australia/2021

Australian Criminal Intelligence Commission . (2018). Annual report. https://www.acic.gov.au/publications/annual-reports/australian-criminal-intelligencecommission-annual-report-2017-18

Australian Criminal Intelligence Commission . (2023). National police reference system. https://www.acic.gov.au/services/national-police-reference-system

10.

Babchishin

K. M.

Hanson

R. K.

Hermann

C. A.

(2010). The characteristics of online sex offenders: A meta-analysis. Sexual Abuse: A Journal of Research and Treatment, 23(1), 92–123. https://doi.org/10.1177/1079063210370708

11.

Babchishin

K. M.

Hanson

R. K.

Vanzuylen

(2015). Online child pornography offenders are different: A meta-analysis of the characteristics of online and offline sex offenders against children. Archives of Sexual Behavior, 44(1), 45–66. https://doi.org/10.1007/s10508-014-0270-x

12.

Bleakley

(2019). Watching the watchers: Taskforce Argos and the evidentiary issues involved with infiltrating Dark Web child exploitation networks. The Police Journal: Theory, Practice and Principles, 92(3), 221–236. https://doi.org/10.1177/0032258X18801409

13.

Bourke

Hernandez

(2009). The “Butner Study” redux: A report on the incidence of hands-on child victimization by child pornography offenders. Journal of Family Violence, 24(3), 183–191. https://doi.org/10.1007/s10896-008-9219-y

14.

Brown

Bricknell

(2018). What is the profile of child exploitation material offenders? In Trends and Issues in Crime and Criminal Justice, Vol. 564. Australian Institute of Criminology. https://www.aic.gov.au/publications/tandi/tandi564

15.

Bursztein

Clarke

DeLaune

Eliff

D. M.

Hsu

Olson

Shehan

Thakur

Thomas

Bright

(2019). Rethinking the detection of child sexual abuse imagery on the internet [Paper presentation]. The World Wide Web Conference, 2601–2607. https://doi.org/10.1145/3308558.3313482

16.

Cale

Holt

Leclerc

Singh

Drew

(2021). Crime commission processes in child sexual abuse material production and distribution: A systematic review. In Trends and Issues in Crime and Criminal Justice, Vol. 617. Australian Institute of Criminology. https://doi.org/10.52922/ti04893

17.

Chen

Jardine

Fan Liu

Zhu

J. J. H.

(2022). Seeking anonymity on the Internet: The knowledge accumulation process and global usage of the Tor network. New Media & Society, 26(2), 1074–1095. https://doi.org/10.1177/14614448211072201

18.

Coen

Napier

(2022). Child sexual abuse material and end-to-end encryption on social media platforms: An overview. In Trends & Issues in Crime and Criminal Justice, Vol. 653. Australian Institute of Criminology. https://doi.org/10.52922/ti78634

19.

Crimes Amendment (Sexual Offences and Other Matters) Act 2014 (Vic) . https://www.legislation.vic.gov.au/as-made/acts/crimes-amendment-sexual-offences-andother-matters-act-2014

20.

Criminal Code Act 1995 (Cth) . https://www.legislation.gov.au/Details/C2019C00043

21.

Crofts

Lee

(2013). Sexting”, children and child pornography. Sydney Law Review, 35(1), 85–106. https://ssrn.com/abstract=2403581.

22.

Cutajar

Mullen

P. E.

Ogloff

J. R. P.

Thomas

S. D.

Wells

D. L.

Spataro

(2010). Psychopathology in a large cohort of sexually abused children followed up to 43 years. Child Abuse & Neglect, 34(11), 813–822. https://doi.org/10.1016/j.chiabu.2010.04.004

23.

DeLisi

Piquero

A. R.

(2011). New frontiers in criminal careers research, 2000-2011: A state-of-the-art review. Journal of Criminal Justice, 39(4), 289–301. https://doi.org/10.1016/j.jcrimjus.2011.05.001

24.

ECPAT International & Interpol . (2018). Towards a global indicator on unidentified victims in child sexual exploitation material. ECPAT International.

25.

Europol . (2020). Exploiting isolation: Offenders and victims of online child sexual abuse during the Covid-19 pandemic. Exploiting Isolation: Offenders and victims of online child sexual abuse during the COVID-19 pandemic | Europol (europa.eu).

26.

Everitt

B. S.

Landau

Leese

(2001). Cluster analysis, 4th edition. Oxford University Press, Inc.

27.

Freeman

N. J.

Sandler

J. C.

(2008). Female and male sex offenders: A comparison of recidivism patterns and risk factors. Journal of Interpersonal Violence, 23(10), 1394–1413. https://doi.org/10.1177/0886260508314304

28.

Gelb

(2007). Recidivism of sex offenders: Research paper. Sentencing Advisory Council. https://www.sentencingcouncil.vic.gov.au/publications/recidivism-sex-offenders-researchpaper

29.

Gottfredson

M. R.

(2006). The empirical status of control theory in criminology. In Cullen

Wright

Blevins

(Eds.), Taking stock: The status of criminological theory (pp. 77–100). Transaction.

30.

Hanson

R. K.

Bussiere

M. T.

(1998). Predicting relapse: A meta-analysis of sexual offender recidivism studies. Journal of Consulting and Clinical Psychology, 66(2), 348–362. https://doi.org/10.1037//0022-006X.66.2.348

31.

Henshaw

Ogloff

J. R. P.

Clough

J. A.

(2018). Demographic, mental health, and offending characteristics of online child exploitation material offenders: A comparison with contact only and dual sexual offenders. Behavioral Sciences & the Law, 36(2), 198–215. https://doi.org/10.1002/bsl.2337

32.

Howard

P. D.

Barnett

G. D.

Mann

R. E.

(2014). Specialization in and within sexual offending in England and Wales. Sexual Abuse: A Journal of Research and Treatment, 26(3), 225–251. https://doi.org/10.1177/1079063213486934

33.

Internet Watch Foundation . (2012). Study of self-generated sexually explicit images & videos featuring young people online. https://www.iwf.org.uk/media/j5wphscw/iwf_study_self_generated_content_online_011112.pdf

34.

Internet Watch Foundation . (2021). The annual report 2021. https://annualreport2021.iwf.org.uk/

35.

Internet Watch Foundation . (2015). Emerging patterns and trends report #1 online-produced sexual content. https://www.iwf.org.uk/media/2saninlk/onlineproduced_sexual_content_report_100315.pdf

36.

Internet Watch Foundation . (2024). What has changed in the AI CSAM landscape? https://www.iwf-ai-csam-report_update-public-jul24v13.pdf

37.

Interpol . (2020). Threats and trends: Child sexual exploitation and abuse: COVID-19 impact. https://www.interpol.int/en/News-and-Events/News/2020/INTERPOL-report-highlightsimpact-of-COVID-19-on-child-sexual-abuse

38.

Jain

A. K.

Dubes

R. C.

(1988). Algorithms for clustering data. Prentice-Hall.

39.

Karriker-Jaffe

K. J.

(2011). Areas of disadvantage: A systematic review of effects of area-level socioeconomic status on substance use outcomes. Drug and Alcohol Review, 30(1), 84–95. https://doi.org/10.1111/j.1465-3362.2010.00191.x

40.

Kaufman

Rousseeuw

P. J.

(1990). Finding groups in data: An introduction to cluster analysis. Wiley.

41.

Lee

Crofts

Salter

Milivojevic

McGovern

(2013). “Let’s get sexting”: Risk, power, sex and criminalisation in the moral domain. International Journal for Crime, Justice and Social Democracy, 2(1), 35–49. https://doi.org/10.5204/ijcjsd.v2i1.89

42.

Levy

(2002). Virtual child pornography: The eroticization of inequality. Ethics and Information Technology, 4(4), 319–323. https://doi.org/10.1023/A:1021372601566

43.

Lievore

(2004). Recidivism of sexual assault offenders: Rates, risk factors and treatment efficacy. Australian Institute of Criminology. https://www.aic.gov.au/publications/archive/archive-150

44.

Martin

(2015). Conceptualizing the harms done to children made the subjects of sexual abuse images online. Child & Youth Services, 36(4), 267–287. https://doi.org/10.1080/0145935X.2015.1092832

45.

Merdian

H. L.

Wilson

Boer

D. P.

(2009). Characteristics of internet sexual offenders: A review. Sexual Abuse in Australia and New Zealand, 2(1), 34–47. https://doi.org/10.1177/1079063210370708

46.

Mitchell

K. J.

Jones

L. M.

Finkelhor

Wolak

(2011). Internet-facilitated commercial sexual exploitation of children: Findings from a nationally representative sample of law enforcement agencies in the United States. Sexual Abuse: A Journal of Research and Treatment, 23(1), 43–71. https://doi.org/10.1177/1079063210374347

47.

Napier

Teunissen

Boxall

(2021). Live streaming of child sexual abuse: An analysis of offender chat logs. In Trends & Issues in Crime and Criminal Justice, Vol. 639. Australian Institute of Criminology. https://doi.org/10.52922/ti78375

48.

Nguyen

T. T.

Nguyen

Q. V. H.

Nguyen

D. T.

Nguyen

D. T.

Huynh-The

Nahavandi

Nguyen

T. T.

Pham

Q.-V.

Nguyen

C. M.

(2022). Deep learning for deepfakes creation and detection: A survey. Computer Vision and Image Understanding, 223(4), Article 103525. https://doi.org/10.1016/j.cviu.2022.103525

49.

Owens

J. N.

Clapp

Craun

S. W.

van der Bruggen

van Balen

van Bunningen

Talens

(2023). Analysis of topic popularity within a child sexual exploitation Tor hidden service. Aggression and Violent Behavior, 68(2), Article 101808. https://doi.org/10.1016/j.avb.2022.101808

50.

Patchin

J. W.

Hinduja

(2020). Sextortion among adolescents: Results from a national survey of U.S. youth. Sexual Abuse: A Journal of Research and Treatment, 32(1), 30–54. https://doi.org/10.1177/1079063218800469

51.

Payne

(2007). Recidivism in Australia: Findings and future research. Australian Institute of Criminology. https://www.aic.gov.au/publications/rpp/rpp80

52.

Quinsey

V. L.

(1986). Men who have sex with children. In Weisstub

D. N.

(Ed.), Law and mental health: International perspectives (pp. 140–172). Pergamon.

53.

Richards

(2011). What makes juvenile offenders different from adult offenders? In Trends & Issues in Crime and Criminal Justice, Vol. 409. https://www.aic.gov.au/publications/tandi/tandi409

54.

Salter

Wong

W. K.

Breckenridge

Scott

Cooper

Peleg

(2021). Production and distribution of child sexual abuse material by parental figures. In Trends & Issues in Crime and Criminal Justice, Vol. 616. Australian Institute of Criminology. https://doi.org/10.52922/ti04916

55.

Seto

Buckman

Dwyer

R. G.

Quayle

(2018). Production and active trading of child sexual exploitation images depicting identified victims. National Center for Missing and Exploited Children. https://www.missingkids.org/content/dam/missingkids/pdfs/ncmecanalysis/ProductionandActiveTradingofCSAM_FullReport_FINAL.pdf

56.

Seto

M. C.

(2019). The motivation-facilitation model of sexual offending. Sexual Abuse: A Journal of Research and Treatment, 31(3), 3–24. https://doi.org/10.1177/1079063217720919

57.

Seto

M. C.

Reeves

Jung

(2010). Explanations given by child pornography offenders for their crimes. Journal of Sexual Aggression, 16(2), 169–180. https://doi.org/10.1080/13552600903572396

58.

Sheehan

Sullivan

(2010). A qualitative analysis of child sex offenders involved in the manufacture of indecent images of children. Journal of Sexual Aggression, 16(2), 143–167. https://doi.org/10.1080/13552601003698644

59.

Steel

C. M. S.

(2024). Artificial intelligence and CSEM – a research agenda. Child Protection and Practice, 2(1), Article 100043. https://doi.org/10.1016/j.chipro.2024.100043

60.

Stiernströmer

Väfors Fritz

Mellgren

Khoshnood

(2022). Demographic characteristics of convicted child sexual abusers in south of Sweden, between 2013 and 2018: A pilot study. Forensic Sciences Research, 7(3), 393–401. https://doi.org/10.1080/20961790.2022.2052590

61.

Tabachnick

B. G.

Fidell

L. S.

(2013). Using multivariate statistics (6th ed.). Pearson.

62.

Thorn . (2014, March 21). Ongoing child pornography distribution via peer-to-peer networks. https://www.thorn.org/blog/child-pornography-distributed-via-peer-to-peer-networks/

63.

Tomak

Weschler

F. S.

Ghahramanlou-Holloway

Virden

Nademin

M. E.

(2009). An empirical study of the personality characteristics of internet sex offenders. Journal of Sexual Aggression, 15(2), 139–148. https://doi.org/10.1080/13552600902823063

64.

U.S. Department of Justice . (2016). The national strategy for child exploitation prevention and interdiction: A report to congress. https://www.justice.gov/psc/national-strategy-childexploitation-prevention-and-interdiction

65.

Victoria Police . (2022). Crime statistics. https://www.police.vic.gov.au/crime-statistics

66.

Wakeling

H. C.

Howard

Barnett

(2011). Comparing the validity of the RM2000 scales and OGRS3 for predicting recidivism by internet sexual offenders. Sexual Abuse: A Journal of Research and Treatment, 23(1), 146–168. https://doi.org/10.1177/1079063210375974

67.

Walker

Hiller

J. E.

(2007). The index of relative socio-economic disadvantage: General population views on indicators used to determine area-based disadvantage. Australian and New Zealand Journal of Public Health, 29(5), 442–447. https://doi.org/10.1111/j.1467-842X.2005.tb00224.x

68.

Ward

J. H.

Jr. (1963). Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association, 58(301), 236–244. https://doi.org/10.1080/01621459.1963.10500845

69.

WeProtect Global Alliance (WPGA) . (2021). Global threat assessment 2021. https://www.weprotect.org/global-threat-assessment-21/

70.

Wolak

Finkelhor

Mitchell

K. J.

(2005). The varieties of child pornography production. In Quayle

Taylor

(Eds.), Viewing child pornography on the Internet: Understanding the offense, managing the offender, helping the victims (pp. 31–48). Wiley. https://doi.org/10.1002/car.916

71.

Wolak

Finkelhor

Mitchell

K. J.

Jones

L. M.

(2011). Arrests for child pornography production: Data at two time points from a national sample of U.S. Law enforcement agencies. Child Maltreatment, 16(3), 184–195. https://doi.org/10.1177/1077559511415837

72.

Wortley

Smallbone

(2014). A criminal careers typology of child sexual abusers. Sexual Abuse: A Journal of Research and Treatment, 26(6), 569–585. https://doi.org/10.1177/1079063213503689

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.21 MB