Does existing legislation on CSEA/CSAM across the Five Eyes nations allow for criminal liability or any other form of accountability with regards to AI-generated CSAM?
Bibliographic record
Abstract
Child sexual exploitation and abuse (CSEA) is considered a violation of children’s rights and dignity (Ngo, 2021). A ‘widespread, worldwide issue of concerning magnitude’ that affects both girls and boys (Simon, Luetzow & Conte, 2020: 2), CSEA may entail a series of negative effects for victims, which can impact their physical, mental or psychological health, their emotional wellbeing, social skills and interpersonal relationships, economic status, as well as vulnerability to future victimisation (Fisher et al., 2017). Within this, technology and related platforms or online environments are considered spaces which can be protective, but equally also raise the risk for children’s safety, increasing their vulnerability to be victimised in CSEA (Simon, Luetzow & Conte, 2020). This vulnerability to victimisation is, in fact, considered to be higher for children than adults (Quayle, 2016). The rapid development of technology has led to the birth of new, immersive forms of technology, which are usually grouped under the umbrella term ‘eXtended Reality’ (XR) (Huang, 2022). Prominent among these emerging technologies is Artificial Intelligence (AI), defined widely by Bahoo, Cucculelli and Qamar (2023: 1) as ‘the system's ability to interpret data and leverages computers and machines to enhance humans' decision-making, problem-solving capabilities, and technology-driven innovativeness’. As such and following the increasing dissemination of child sexual abuse material (CSAM) noticed across the clear and dark web, AI can prove to be a valuable tool in the efforts against CSEA by allowing the invention of detection intelligence algorithms that will use deep-learning technique as methods of accurate detection of CSAM online (Lee et al., 2020; Ngo, McKeever & Thorpe, 2023). However, on the downside, and mainly with regards to its content generative aspect, AI can also be misused by offenders to create CSAM with varying levels of realism that can often be hardly distinguishable from real-life material (Internet Watch Foundation, 2023). Irrespective of whether AI-created CSAM involves artificial children or children modelled after real-life children, there is widespread concern that it can be a pathway to higher levels of CSEA offending that may include the sexual exploitation and abuse of children in real life (Internet Watch Foundation, 2023). As such, it requires a robust and clear legislative response, particularly with regards to the issue of accountability over AI-created CSAM. This call comes amidst a hotly contested debate, with some stakeholders promoting notions that CSAM created via generative AI does not actually hurt real children or that it may also serve to divert potential offenders from sexually exploiting and abusing real children, and others who fear that generative AI-created CSAM may well be the first step on a pathway towards higher offending in CSEA with real children (Internet Watch Foundation, 2023). Based on the above, examining the existing legislative context of the Five Eyes countries, which comprise Australia, Canada, New Zealand, the United Kingdom (UK) and the United States of America (USA), becomes crucial in order to assess the readiness of their regulatory frameworks against phenomena of AI-created CSAM. These countries have been selected due to their democratic and open political systems, their high levels of technological advancement and literacy, as well as their progressive and advanced legislative systems, which often serve as the regulatory blueprints for other countries across the globe who often wish to model their legislation after them.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.012 | 0.027 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.005 | 0.005 |
| Scholarly communication | 0.008 | 0.009 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.013 | 0.010 |
| Insufficient payload (model declined to judge) | 0.030 | 0.008 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".