The question of whether autistic individuals might be predisposed to engaging with computing systems has its origins in work by Baron-Cohen and colleagues in 1997, who highlighted that working with objects and systems rather than people may be reflective of the broad autistic phenotype. This idea gained traction in both the media and among law enforcement personnel, with organizations such as the UK’s National Crime Agency suggesting that autistic individuals may have a greater propensity for cybercrime. However, attempts to test this relationship empirically have produced deeply conflicting results over the past decade, with different studies pointing to autism diagnosis, autistic traits, or neither as the relevant predictor.

Previous research has been marked by methodological differences that may explain some of the contradictory findings. In a 2015 survey of 296 undergraduate students, Seigfried-Spellar and colleagues measured computer deviant behavior and autistic traits, finding that 60 percent of participants reported some form of computer deviant behavior and that those who had engaged in virus writing, identity theft, or cyberbullying recorded significantly higher scores on the Autism Spectrum Quotient than students who had not. However, there was no difference in level of autistic traits between students who had engaged in hacking and those who had not. In a more recent study, Payne and colleagues sampled 290 internet users and found that autistic traits, but not an autism diagnosis, were associated with higher engagement in cyber-dependent crime, while an autism diagnosis was related to a reduced rather than an increased involvement in such crime.

Yet contradictory findings were subsequently reported by Lim and colleagues in 2023, who found that an autism diagnosis, but not autistic traits, predicted engagement in cybercrime in a sample of 302 participants. Cyber-dependent crime was reported by 20 percent of the autistic sample and 11.2 percent of the non-autistic sample. More recently, Maras and colleagues examined the relationship between autism diagnosis and cybercrime among adolescents in the UK Millennium Cohort Study, a longitudinal study of over 19,000 individuals, and found that after controlling for demographic variables, cognitive ability, family background, and environment, autism diagnosis was not a significant predictor of cybercrime involvement. These inconsistencies suggested that methodological limitations, rather than genuine differences in the underlying relationship, might be responsible for the conflicting data patterns.

To address these limitations, Lim and Brewer conducted a partial replication of the 2023 study with a much larger sample of autistic adults. The researchers recruited 344 participants through the online crowdsourcing platform Prolific, of whom 138 reported having a formal diagnosis of autism from a qualified professional and 206 confirmed they had not received such a diagnosis. Participants ranged in age from 18 to 75 years. The sample was substantially larger than the autistic samples in previous studies, which had included only 25 and 23 autistic participants respectively, providing increased statistical power and variance on the key measures.

The study incorporated several methodological improvements designed to address the weaknesses of prior research. Participants who endorsed engagement in cybercrime were asked to provide a brief description of what they did and why, ensuring that any disclosure of cybercrime was in fact actual criminal activity. Responses that were clearly not cybercrime, such as accessing a family member’s email account with their permission or learning to hack a server as part of IT training, were modified to reflect no engagement. The researchers also included two measures to control for differences in participants’ general tendencies to lie or to respond in self-enhancing ways: the Lying in Everyday Situations Scale and the Balanced Inventory of Desirable Responding Short Form. Additionally, they measured autistic traits using the 50-item Autism Spectrum Quotient rather than a shorter version, allowing examination of whether different types of autistic traits were more predictive of cybercriminal activity than others.

The results revealed that individuals with an autism diagnosis were 7.30 times more likely to report engagement in cyber-dependent crime than individuals without a diagnosis, although the researchers noted that the wide confidence interval indicated this estimate was imprecise. In contrast, autistic individuals were not significantly more likely to report engagement in cybercrime more broadly, which included both cyber-dependent and cyber-enabled crime. Advanced digital skills was also a significant predictor of engagement in cyber-dependent crime, but the odds ratio values for this relationship were small. Age, gender, autistic traits, and theory of mind did not significantly predict engagement in cybercrime. A mediation analysis revealed that the relationship between autism diagnosis and cyber-dependent crime was not mediated by advanced digital skills or theory of mind, suggesting that the pathway through which autism diagnosis influences cyber-dependent crime remains unclear.

The researchers conducted exploratory analyses to examine whether differences in self-reported cyber-criminality between autistic and non-autistic participants were due to differences in self-disclosure rather than actual differences in criminal behavior. The results showed that neither group displayed a propensity for lying or socially desirable responding, no group differences were found on either measure, and autism diagnosis remained a significant predictor of cyber-dependent crime after controlling for scores on the lying and social desirability measures. However, the exploratory analyses also revealed that autistic individuals who reported engagement in cybercrime had a higher propensity to lie and were less likely to respond in a socially desirable manner than non-autistic individuals who reported engagement in cybercrime, raising questions about the reliability of self-report data in this research area.

The findings suggest that involvement in cyber-dependent crime may be more prevalent among autistic than non-autistic adults, but the researchers caution that it would be premature to conclude that autism or autistic traits per se are risk factors for engagement in cybercriminal activity. The complex multi-factor predictors identified in the UK Millennium Cohort Study, including being male, domestic violence between parents, low parental monitoring, low wellbeing, self-harm, exclusion from school, and spending more time online gaming, highlight the importance of developing a more sophisticated understanding of the environmental and individual difference variables that might exacerbate cybercrime risk for some autistic individuals. Future research will need to focus on whether there are specific characteristics of autistic individuals that might render them vulnerable to becoming involved in cybercrime, and to corroborate the current findings using objective data rather than self-report measures alone.

Subject of Research: Cyber Dependent Crime May Be Predicted by Autism Diagnosis but Not Autistic Traits

Article Title: Cyber Dependent Crime May Be Predicted by Autism Diagnosis but Not Autistic Traits

Article References: Lim, A., & Brewer, N. (2026). Cyber Dependent Crime May Be Predicted by Autism Diagnosis but Not Autistic Traits. Journal of Autism and Developmental Disorders. https://doi.org/10.1007/s10803-026-07580-1

Image Credits: AI Generated

DOI: 10.1007/s10803-026-07580-1

Keywords: Cyber, Dependent, Crime, Predicted, Autism, Diagnosis, Autistic, Traits, scientific research