To know or not to know: Should crimes regarding photographs of their child sexual abuse be disclosed to now-adult,unknowing victims?

Introduction

Imagine the following scenario. During their investigation of crimes related to abusive images, law enforcement agents (LEAs) successfully identify the very young victim – a toddler – who features in the images. The images are historic, and the victim is now an adult. There is no evidence to suggest the victim is aware of the images and, because of their very young age when the crimes were committed, they may remain unaware of the actual abuse. Now imagine that victim is you. Would you want to be informed of the crimes and the existence of the images?

The identification of children who feature in abusive images (AI) has always posed real challenges (Holland, 2005; Carr and Hilton, 2011). ¹ Victim identification has been coordinated nationally, but only by one underfunded unit within the Child Exploitation and Online Protection Centre in the United Kingdom (UK) (Gillespie, 2011: 332–333) and, in 2014, the National Society for the Prevention of Cruelty to Children claimed that police were ‘overwhelmed’ by the amount of AI collected from seized computers (Conway, 2014). Although victim identification remains far from straightforward, 2014 saw the launch of the national Child Abuse Images Database (CAID), a system designed to speed up the process of cataloguing and analysing images (Home Office, 2018). All police forces now have access to CAID and there are clear indications that it has increased the number of victims being identified and speeded up the identification process (Home Office, 2018). The rolling out of CAID and greater success in terms of victim identification are significant positive developments but, at the same time, they have increased the likelihood of the dilemma that is the concern of this paper. For where victims are successfully identified and located, this raises difficult ethical issues pertaining to whether to disclose the existence of AI to them.

We note at the outset that this paper is not focused on cases where the identified victims are still children. For such victims, there are significant and urgent matters that must be addressed following their identification and location, such as ascertaining whether they are now in a safe environment and whether they have received counselling support, matters beyond the scope of this paper (Palmer, 2005). ² Rather, our focus is on images depicting historic child sexual abuse where the identified victim is now an adult. Where the images depict historic crimes, four different scenarios can be envisaged. First, LEAs may identify a victim who is aware that her abuse was recorded in photographs and that these photographs were disseminated (V1). ³ Secondly, they might identify a victim who is aware of the AI, but not that they were distributed to others (V2). In the third possible scenario, they may identify a victim who is aware of her sexual abuse but not of the AI because they were taken covertly (V3). Finally, LEAs could be faced with the scenario with which we began this paper: they could identify a victim who was abused at a very young age and is thus unaware both of the sexual abuse and the AI (V4), as might occur, for example, when unaware toddlers or pre-school victims become adults (Anon, 2010; Morris, 2010). To complicate matters further, it is highly likely that, when they identify a victim featured in a photograph, LEAs will not know which of these scenarios they face unless they have identified a victim already known to them and are aware of the degree of her knowledge regarding the images.

In each of these scenarios, there is a substantial risk that disclosing the abuse and/or existence of the AI for V3 and 4, or the fact that they have been accessed by others for V1 and 2, will cause the now-adult victim (further) harm. Although still in its infancy, research involving AI victims and the counsellors and therapists who treat them has revealed that, besides the trauma caused by the actual abuse, AI victims suffer additional, unique psychological harm if they are aware of the existence of AI and that these images have been distributed on the internet (Gewirtz-Meydan et al., 2018; Martin, 2015; 2016; von Weiler et al., 2010; Lindauer et al., 2014). The availability of the AI for others to view means that achieving ‘closure’ for victims can be impossible. Leonard’s research has shown that ‘we cannot approach [these victims] as post trauma as they are still very much living and experiencing the trauma…at any time, on any day, in any country someone…could be looking at the pictures of them and using them as sexual stimulus for their own self-gratification’ (Leonard, 2010: 254).

Even in the case of V1, who is already aware that the AI exist and have been distributed, informing her that the AI have been found in the possession of another offender who has downloaded them from the internet, for example, could cause her additional distress by reminding her again of her abuse and that the images are still ‘out there’ for others to access (Leonard, 2010). Yet at the same time, it could be argued that victims have an interest in knowing this information pertaining to themselves (Palmer, 2005: 65) and a failure to disclose would deny them the opportunity to take self-ownership of this experience and deal with it. Thus, the crucial question is whether to disclose and this may depend, in part, on the category of victim that LEAs face. There is almost no research on how to resolve this dilemma, which also poses difficulties for counsellors treating unknowing AI victims (Martin, 2015; 2016). Palmer (2005: 65) has noted that ‘[w]e know little about the impact of such a revelation on the now adult “child victim” and need to think strategically about why and when such a disclosure may be necessary’. Moreover, Taylor and Quayle (2003: 207) have expressed the view that police policy guidelines on the identification of now older victims are ‘poorly thought through and expressed’.

To begin to address this paucity of research, this paper’s unique contribution is to provide an ethical, theoretical analysis of the issues and frame this analysis (and a suggested way forward) around the victim’s interest in knowing information pertaining to herself and her history, and her interest in not knowing this information. To the best of our knowledge, this is an issue that has not been explored in the victimology and criminological literature. Our analysis should be of significance beyond the particular AI context, because it will resonate with other situations in which victims are unaware of crimes committed against them (assuming that information about the crime subsequently comes to the attention of the authorities). Take, for instance, a victim who is raped while unconscious and remains unaware of the rape (Anon, 2015; 2016; Gardner, 2007); victims of a gynaecologist who takes explicit photographs during pelvic examinations without their knowledge (Pearce, 2014; Klosterman, 2014); unaware victims of ‘revenge pornography’ (Gillespie, 2015); and unwitting victims of a fake charity scam fraud (Button et al., 2009). This also suggests that there are important implications for the victimology literature: is there a need to consider another typology of crime victim – the unaware or unknowing victim – alongside existing conceptions of, for example, the ideal and non-ideal victim (Christie, 1986), the innocent victim (Mendelsohn, 1963) and the hidden victim (Chakraborti and Garland, 2012)? This little-explored ⁴ category of victim raises numerous issues that merit consideration for future research – such as, for instance, whether it should also encompass victims who do not construe another’s behaviour towards them as a criminal offence (Mooney and Ost, 2013) – alongside the disclosure dilemma that is the focus on this paper.

This paper unfolds as follows. First, because this disclosure dilemma has not been addressed in victimology or criminological debates, we look to literature in another academic discipline to inform our analysis. A sustained exploration of the so-called right to know and not to know personal and sensitive information concerning oneself has occurred in healthcare ethics and law (HCE&L) literature regarding the processing of genetic information (Chadwick et al., 2014; Laurie, 2002; Harris and Keywood, 2001). ⁵ Our purpose is not to claim that the genetic information context and the disclosure dilemma regarding AI victims are identical, for they are not; there are different issues at stake. In particular, there is a clear contrast in terms of the content of the information – information regarding an individual’s health versus information concerning the existence and distribution of a photographic record of a sexual crime committed against a person when she was a child. This gives rise to distinct psychological ramifications. Moreover, given that it is often a child’s family member or other trusted adult who creates the AI (Gewirtz-Meydan et al., 2018: 242), disclosure may well have different ramifications for family relationships than has the revelation of genetic information pertaining to a person’s health, as we shall discuss. However, in both cases, it is personal and sensitive information, the disclosure of which could be life changing. As we shall see, reflecting on the analysis in the HCE&L literature and the similarities and differences between the two contexts can help inform an appropriate conceptualisation of the issues in the context of unknowing AI victims, which can, in turn, assist in finding a way to resolving the disclosure dilemma.

Secondly, we reveal the difficulties created if the issue of informing unaware victims of crimes committed against them is conceptualised as rights based. Instead, we argue that the matter is best viewed through the lens of interests in knowing and not knowing, because such a conceptualisation enables a more constructive balancing assessment of the issues at stake. Thirdly, informed by the theoretical analysis in the genetic information debate, we argue that the interests at stake are grounded in autonomy and/or spatial privacy (the latter being a form of psychological separateness or state of not knowing). To find a way to resolve the disclosure dilemma, these interests must be considered alongside the consequences of disclosure: revealing information regarding AI (and sexual abuse for V4) could positively empower now-adult victims but could well cause them further damage by way of serious psychological harm and suffering. The way in which these interests and concerns can be balanced is addressed in our penultimate section, where we evaluate the arguments for and against disclosure for the different categories of AI victims. Finally, we draw the elements of our analysis together to formulate a proposed way forward to resolve the disclosure dilemma regarding unknowing AI victims, and consider the implications for victimology.

A right to know and not to know in HCE&L (genetic information) and its possible application to the AI context

The right to know and not to know genetic information first began to be debated in the 1990s following significant developments in genetic screening (Chadwick et al., 2014). In some cases, such screening can provide certain knowledge that an individual will develop condition X; in others, the person’s genetic status will provide less precise information about the percentage risk they will develop condition Y. These opportunities to undergo genetic testing for late-onset disorders such as Huntington’s disease, ⁶ for instance, have given rise to numerous ethical dilemmas pertaining to disclosure. For our purposes, the first of the most germane of these relates to the fact that some individuals at risk of having such a disease – because, for example, one of their parents has been diagnosed with the condition – would wish to be tested to gain knowledge about whether they will develop the disease and to structure their lives accordingly. If tested, these individuals could claim a right to know the results of the genetic screening. Conversely, someone may initially choose to be tested but then decide they would prefer not to know the results rather than take the risk of being given a devastating, life-changing diagnosis. The second pertinent dilemma arises where the particular disease is hereditary and a late-onset single gene disorder, such as Huntington’s disease. If one family member is screened and the result is positive, it is 100% certain that one of their parents will have the gene and there is a 50% chance that each of their siblings will (Laurie, 2014b). This test will thereby indicate the genetic status of these individuals (because Huntington’s disease is late onset, they might not have symptoms even if they have the dominant gene), but they may be completely unaware their relative has had the genetic test. Should they have a right to know about this test and its outcome? And should they have a right not to know the outcome?

There is a noteworthy parallel here relating to the contrasting levels of knowledge possessed by the various parties in the genetic and AI contexts. The AI disclosure dilemma that we have introduced involves a spectrum of victims, some of whom may have incomplete knowledge of the existence of the personal, sensitive information concerning crimes committed against them, or no knowledge whatsoever. Take, for instance, AI V1 (who knows her abuse was recorded and that the AI were then distributed) and the person who requests genetic testing: both are aware that the information does or could exist. In contrast, relatives who are unaware of the genetic test and its implications for their health and AI V3 and 4 possess a lack of knowledge of the information’s existence. The same is true for V2 if the information relates to the distribution of the AI. This puts the person who requests genetic testing and AI V1 in a different position to unaware relatives and AI V2–4.

What conceptual grounding exists for considering whether these different parties have a right to know or not to know the information in question? In the HCE&L literature, the discourse of rights to know and not to know tends to be grounded in the interests of autonomy and privacy (Borry et al., 2014; Chico, 2015). Looking first to autonomy and working with the definition of autonomy as ‘self-rule that is free from both controlling interference by others and from certain limitations such as inadequate understanding that prevent meaningful choice’ (Beauchamp and Childress, 2013: 101), there is an obvious connection with the right to know. For if there is significant information pertaining to A’s health and/or medical treatment that B possesses and denies A the knowledge of, then B prevents A from making meaningful choices relating to her self-governance. But the grounding of this right in autonomy has been contested where the individual has not previously expressed a wish to know the information. Husted (2014) has cautioned that we should be wary of conceiving autonomy as the basis for the right to know in cases where the disclosure is unsolicited, because this leads to both an apparent denial and enhancement of autonomy. Autonomy might seem to be denied, in that disclosing information deprives the individual of her autonomy because the decision is made by the discloser. However, this disclosure is made for the purposes of enhancing the individual’s autonomy by providing her with significant information that she should have available to her as the decision maker in her life (Husted, 2014). Yet, ultimately, Husted goes on to conclude that what might seem to be an enhancement of autonomy actually leads to ‘morally forced choices’ (Husted, 2014: 36), because being informed that one is the carrier of a genetic condition that brings with it serious health problems could lead to an individual turning away from the life choices they have previously made. They may decide, for example, not to have children, or to avoid any serious emotional involvement with others.

This ambiguity suggests that autonomy might not be as secure a foundation for the right to know as it initially seems, unless the individual has expressed a clear prior wish to know the information in question and has exercised a meaningful choice regarding it (Keren-Paz, 2017: 422). Although such a prior wish will exist in the genetic testing context where the individual has requested that tests be carried out (provided she does not subsequently decide that she no longer wishes to know), it will not exist where a relative is unaware that a family member has undergone a genetic test, the results of which have serious implications for her health. Nor will it exist for unknowing AI victims. In other words, for autonomy to be relevant, the individual must be aware there is something to know. This argument might be challenged by the existence of a growing body of empirical research indicating that, when asked whether they would wish to be told about genetic information that they were unaware of, a significant number of study participants would wish to be told (Wolff et al., 2007; Middleton et al., 2016; Heaton and Chico, 2016; Dheensa et al., 2016). Notably, it was especially in the case of genetic information about a condition that steps can be taken to avoid, or where treatment is available, that disclosure of such previously unknown information was favoured (Heaton and Chico, 2016: 114; Middleton et al., 2016: 23). However, fewer participants state that they would wish to be informed of a fatal genetic condition for which there is no treatment (Dheensa et al., 2016: 294), and some would not wish to know regardless of how serious the non-preventable condition is (Heaton and Chico, 2016: 115). It should be borne in mind that these studies only provide hypothetical views; that is, participants were asked what their views would be in supposititious circumstances and ‘it is impossible to know how closely these [hypothetical views and views in a real-life situation] are aligned’ (Middleton et al., 2016: 28; Heaton and Chico, 2016: 118; Dheensa et al., 2016: 291). Also, the fact that not all participants would wish to be informed means it cannot be assumed that disclosure of unknown genetic information would support autonomy, for such an assumption would fail to take account of the ‘beliefs and preferences of [each] individual’ (Middleton et al., 2016: 28). Only a prior expressed wish to know the information in question would reveal a particular individual’s autonomous views accurately. Moreover, no such empirical research, hypothetically based (such as involving the scenario with which we began this paper, for instance) or otherwise, exists regarding views on disclosing the existence of AI to our knowledge.

Looking next at the right not to know, although some authors conceive that this right emerges from the interest of autonomy, its connection with autonomy is also disputed. Certainly, if a right not to know were assumed to exist as a default position, then this would seem to give rise to a paternalistic approach of protecting individuals from knowing information even though they have no idea of the information’s existence or its content, which would hardly seem to be autonomy enhancing. However, there appears to be greater synergy between a right not to know and autonomy if we focus on the situation where an individual expresses a wish not to know in advance. Andorno argues that the right embraces autonomy because the decision not to know is made by the patient who has been tested – she exercise a ‘right to informational self determination’ (Andorno, 2004: 436–437; Strasser, 1986; Knoppers, 2014). This grounding is strongly disputed by Harris and Keywood (2001: 421), who perceive a right not to know as being a claimed right to remain in ignorance. Viewed in this light, they argue that ‘where the individual is ignorant of information that bears upon rational life choices she is not in a position to be self-governing’.

The crux of the matter for Harris and Keywood appears to be that an individual can only make an autonomous decision if she avails herself of crucial information necessary to make rational life-choices for the future. Similarly, Ost has stated that asserting a right not to know is ‘misguided rhetoric’ because a decision to refuse to hear information that relates to one’s medical condition and treatment is irrational, and ‘irrationality is an autonomy-defeating condition’ (Ost, 1984: 304, 306). Although we have limited space to critique this position, we would question why it should be the case that an individual cannot make an autonomous choice to refuse to hear information to protect her psychological integrity. Granted, she may be depriving herself of knowledge that could alter certain life choices she is making that she considers to be rational, and some of these choices may be self-defeating (if, say, she is only likely to live another five years rather than 25 (Harris and Keywood, 2001)). But does this fail to satisfy the aforementioned definition of autonomy? Can it not be a meaningful choice based on sufficient understanding, provided she has taken into account the advantages and disadvantages of knowing the results of the genetic test? Why should her choice to remain unaware of this information to protect her psychological integrity be irrational? And, given the potential for serious harm that we go on to discuss, it is even more difficult to argue that a knowing AI victim who chooses not to be informed of any new cases involving her AI makes an irrational choice.

When it comes to unsolicited disclosures, Laurie has proposed an alternative grounding of the right or interest of the person to whom the information relates, which could better capture the value at stake than autonomy (we will explore the distinction between ‘right’ and ‘interest’ in a later section). For, ‘[i]f I do not know that there is something to know, then I cannot meaningfully exercise a choice [to know or] not to know’ (Laurie, 2014b: 56; and see Andorno, 2004: 437). Returning to and critiquing Husted’s earlier contention, how can an unsolicited disclosure cause an infringement of autonomy given that the person informed ‘has never exercised their autonomy over the said information’ (Laurie, 2014b: 56)? However, although it may not be autonomy that is being encroached upon through an unsolicited disclosure, a violation could still occur. In our view, this violation relates to the person’s spatial privacy, which Laurie explains in the following way:

A realm of psychological spatial privacy allows us to create and retain a greater degree of control over our own concept of who we are and how we choose to engage with the world…Control of information about ourselves must be an essential part of any concept of ourselves as autonomous persons…[including] the facility not to accept the information ab initio. A concept of ‘control’ which is wide enough to encompass this notion permits us to retain a private sphere that is truly our own. Furthermore, it allows us to maintain that unsolicited revelations of personal information [are] an invasion of that sphere, even when such revelations about ourselves are made to ourselves. (1999: 119, 124)

If the basis for a right or interest in not knowing is construed as residing in privacy as suggested above, such an analysis can offer a neutral starting point to address disclosure dilemmas. This is because entering another’s state of privacy through an unsolicited disclosure can sometimes be justified, but, on other occasions, it can constitute ‘unwarranted trespass’ (Laurie, 2014b: 59), depending on the factors for and against disclosure in the particular case. It is here that the significance of Millian consequentialist considerations becomes apparent (Mill, 1993), because a particular factor requiring consideration would be the potential benefit or harm that disclosure may cause to the person.

The initial point to consider is that information regarding an unwanted genetic status pertaining to a serious, non-treatable or preventable condition and information regarding AI images is disruptive. Although it may be empowering, enabling self-ownership of significant personal information, knowledge of this information potentially alters the person’s sense of self in a negative way (Turner, 2009); knowing it is likely to cause distress and harm. Unsolicited disclosures invade individuals’ psychological spatial privacy with knowledge that could have consequent serious adverse effects upon them. In cases of single gene disorders such as Huntington’s disease, the future of the person whose genetic status is revealed could be set in stone in the sense that she will develop condition X and might die as a result, and knowledge of the historic abuse and/or the existence and dissemination of images could cause the AI victim to experience considerable long-term psychological suffering with no certain end point. ⁷ Although the exact duration and severity of the psychological suffering caused to different AI victims and different individuals whose genetic predisposition to a particular condition is revealed could vary, nonetheless it is difficult to conceive of cases where serious psychological suffering would not be caused, unless the condition revealed by genetic testing is preventable or manageable with treatment. Moreover, once this information is known, neither the individual whose genetic status is revealed, nor the AI victim, can do anything about it; although both parties could receive counselling support to help them find a way to deal with the knowledge, they cannot alter the genetic information nor the fact that the abuse and the recording of this abuse has occurred. Neither party can ‘unknow’ nor ‘intentionally abandon’ (Turner, 2009: 364) the information, and knowing it can be life changing, albeit that information regarding AI relates to the past and genetic information to the future.

There are, however, notable distinctions between the potential effects of disclosure in the genetic and AI information contexts. First, in cases where the information pertains to a genetic predisposition to a treatable or preventable condition, the disclosure could enable the individual to take action to reduce the risk of developing the particular disease, or actually prevent the genetic mutation from manifesting; the future harm could be avoided (Laurie, 2004; Takala, 2001), thereby having a greater positive (preparedness) effect than a negative one. Indeed, the empirical studies referred to earlier suggest that where action can be taken to prevent the future harm from occurring, this is the strongest factor influencing participants who expressed a wish to be informed of unknown genetic information pertaining to a condition that would be detrimental to their health (Heaton and Chico, 2016: 114). Thus, although disclosing a person’s genetic status can, in itself, be harmful if the condition is serious and non-treatable or preventable, it can be significantly beneficial where the condition is treatable or preventable – disclosure can avoid the harm of the manifestation of the genetic risk. For AI victims, it is difficult to see how disclosure could enable any preventative action (because the harm has already occurred) or bring about any form of relief. Yet there might be a positive outcome in terms of empowerment, as we will discuss in our penultimate section. In both contexts, therefore, an unsolicited disclosure of the information could be an incursion into spatial privacy that has beneficial effects that could justify the incursion, but these beneficial effects are likely to be very different and of differing value.

Secondly, there are distinct differences regarding the psychological and relational ramifications of disclosure. Research demonstrates that child sexual abuse victims suffer enduring physical and mental harm (Finkelhor, 1984), and the existence of an AI can serve ‘to perpetuate the images and the memory of [the] abuse’ (Taylor and Quayle, 2003: 24). AI available online can feature abuse involving a child performing sexual acts upon adults or other children, adults having sexual intercourse with the child, and abuse that is even more severe, including bestiality, torture and degrading acts (Taylor and Quayle, 2003: 32). It should therefore be unsurprising that AI victims suffer ‘from feelings of shame, hate and disgust or loathing’, and have a ‘higher susceptibility to post-traumatic stress disorder, depression and psychoses’ (von Weiler et al., 2010: 214, 217). In short, ‘the existence of abusive images always [leads] to additional psychological stress’ (von Weiler et al., 2010: 216; Gewirtz-Meydan et al., 2018). And where the abuser and creator of the AI is a family member, as is commonly the case (Gewirtz-Meydan et al., 2018: 242), abuse may be a long-held family secret. Disclosure of this abuse and the existence of AI will have huge, negative consequences, potentially for all family members. Furthermore, some relatives may take the side of the abuser, with the consequence that the family is torn apart (Münzer et al., 2016). In contrast, a person who learns of genetic information with serious implications for her health may turn to her family for support, albeit that the existence of such information can also be divisive and cause other family members to suffer mental harm (Dheensa et al., 2016: 298; Gargiulo et al., 2009). Awareness of the information, especially where the condition is fatal and non-treatable or preventable, is likely to cause serious mental distress, anxiety and depression, but is unlikely to give rise to any feelings of guilt or shame.

Thirdly, in genetic testing, the harm of the manifestation of the genetic risk either might, or will, befall the individual despite whether they are notified in advance of their genetic status. In contrast, for unaware AI victims, it is the notification itself that creates the likely harm. ⁸ We accept that for V1 and V2, their awareness of the existence of photographs of their abuse may mean they have an awareness of a risk that these images could be found in the possession of others, but the certainty of this is only confirmed by the disclosure, which is thereby apt to generate the harm. Thirdly, the knowledge that comes with disclosure in the AI context is black and white – the abuse has occurred, and photographs exist and have been distributed. In many genetic status cases, however, the knowledge will not be as precise. Other than in single gene inherited disorders such as Huntingdon’s disease or cystic fibrosis, there is likely to be a percentage consideration to weigh up (Heaton and Chico, 2016); although it may be that an inherited gene sequence increases the chance of acquiring a particular condition, whether the individual goes on to develop the disease is not certain and depends on a complex group of factors, of which genes are just one component.

The similarities and distinctions between the genetic and AI information contexts resonate. Both disclosure contexts involve differing levels of knowledge, the real risk of harm, and the inability to unknow disruptive information that give rise to morally significant autonomy and spatial privacy concerns. The distinctions, particularly the lack of any positive harm-avoiding effects of disclosure for AI victims, provide strong prima facie grounds to suggest that a careful, precautionary approach should be taken to the matter of whether to disclose the existence of images to AI victims. However, before exploring this further, the next question to be addressed is whether a victim’s lack of knowledge of the existence and/or distribution of AI images means that recognising any rights to know and not to know would pose an intractable conflict.

The dilemma of respecting AI victims’ ‘rights’ without one thwarting the other

The significant matter that we have alluded to concerning foreknowledge of the information’s existence means that although there is usually no conflict between the right to know and not to know for the person who has undergone genetic testing, and this could also be true for AI V1, there is a substantive conflict between these putative rights for unaware relatives in the genetic context and AI V2–4. The person who has undergone genetic testing is already aware there is sensitive information about herself that she may or may not wish to know. Taking into account the reasons in favour of knowing and against knowing, she can then exercise her autonomy in deciding whether to know the results. Similarly, because she is aware of the existence of the AI and their accessibility to others, the same is true for V1 provided she has been given the opportunity to decide whether she wishes to be informed of future cases involving her images. She can make an autonomous decision, informed by her past experiences and awareness of the earlier crimes. If V1 has not been presented with the opportunity to make this advance decision, however, we are faced with the same problem that we will now discuss regarding relatives who are unaware of the genetic tests and AI V2–4.

Unaware relatives cannot make an autonomous decision to know or not to know the results of the test without first being informed that their relative has been tested. Likewise, V2–4 have no knowledge that information exists concerning crimes committed against them (that is, crimes related to distribution and accessing images for V2, and crimes related to abuse, creation, distribution and accessing of the AI for V3 and 4). They are, therefore, unable to exercise their autonomy and weigh up the reasons for and against knowing this information without first being informed of its existence by LEAs. Thus, to exercise a right to know, V2–4 need to be told that information is available about themselves and it is difficult to envisage how they could be alerted to this information’s existence without its essence being revealed and, thereby, their right not to know being thwarted.

Revealing the essence of the information might be avoided in the case of unaware relatives in the genetic context. They could, for instance, be informed that there is certain genetic information that may impact on their health without the specifics of this information being disclosed, and then take the decision as to whether they wish to know more, thereby enabling them to take preventative measures where possible (Suthers et al., 2006: 670). Likewise, a similar means of disclosing partial facts to AI V2–3 could be conceived of; they might be told that further information is available pertaining to sexual abuse crimes committed against them and this would not, in itself, reveal the essence of this information. However, this is arguably tantamount to knowing; if we take seriously the right in not knowing, then this right might be encroached upon if victims are made to know something at all. ⁹ And disclosure of such a vague statement could still lead to V2–3 guessing the nature of information, without having any similar opportunities available to enable them to avoid harm.

What is more, it is even harder to see how the information’s existence could be disclosed without revealing its essence for V1 (where she has not been given an opportunity to decide whether she wishes to be informed of future cases involving her AI) and V4. V1’s pre-existing knowledge of the existence and distribution of the AI means it is highly likely she will realise the ‘further information’ concerns other crimes related to these images. V4 knows nothing of the abuse or the existence of the AI and so any disclosure would have to be couched in extremely vague terms (such as ‘there is certain information and evidence relating to crimes committed against you when you were a young child’) to avoid any revelation of the nature of the crimes committed. Such a nebulous disclosure would preclude any meaningful exercise of V4’s autonomy in deciding whether she wishes to know or not to know this information. Giving V4 the information required to enable her to exercise her right to know is far more likely to defeat her right not to know, thereby thwarting her ability to reject personal information about herself.

If, however, LEAs conclude that keeping AI victims in the dark is the only way to avoid any risk of disclosing the essence of the information to them, they deny them the right to know the information. How, then, are the rights to know and not to know to be reconciled? To answer this question, we may have to move away from the language of rights.

Rights or interests?

The language of rights is often used as convenient shorthand when it comes to matters relating to standards of human behaviour towards others and the protection that should be afforded to the fundamental rights of all human beings. The matters at stake in the dilemma that concerns this paper would, at first glance, appear to be appropriately conceptualised as rights based because they relate to the way in which LEAs should act towards victims of crime and information that is fundamentally private and personal (despite being related to criminal activity). But it is necessary to take a step back and assess whether it is in fact helpful to conceptualise this disclosure dilemma as one involving rights, or whether it is more fruitful to conceptualise the issue in terms of interests.

Although it may be controversial to go so far as to argue that rights are derived from moral principles and are ends in themselves (but see Peerenboom, 1995, 361–362), rights commonly reflect fundamental values and command a prima facie ground for respect (Dworkin, 1984). In contrast, although interests can be conceptualised as ‘the anchor of rights’ (Scanlon, 2009: 76), whether an individual’s particular interest should be protected depends on consequential considerations and the balancing of this interest against any competing interests (Scanlon, 2009: 78). And notably, although the onus is placed on those who would impinge on a right to justify this in the strongest terms and by the least impactful means, there are no prima facie duties as such – or at least there are softer obligations – to pay due recognition to interests in weighing up what ought to be done. ¹⁰

Recognition of rights in law has particular advantages for their protection, because the law can impose a duty to respect the right in question and demand a remedy on its violation. The conceptualisation of the disclosure issue as involving rights in the genetic information debate is understandable given the dominance of patient autonomy and rights in the contemporary professional, ethical and legal models of healthcare. Notably, in a broader healthcare context, the right to know has been given legal authority through the now-recognised prudent patient test: the patient has a right to know material information necessary to make an informed decision regarding their healthcare. The doctor’s duty to respect this right is recognised, and dereliction of this duty that causes harm is actionable through the tort of negligence. ¹¹ Of more specific pertinence, the potential application of human rights law to genetic information offers some interesting insight into how law could come into play in disclosure contexts. A right to know the result of genetic screening pertaining to one’s own health could be claimed under Article 8 (which protects private and family life and autonomy) ¹² of the European Convention on Human Rights (ECHR). Gilbar and Foster (2016: 121) have argued this would be framed as concerning ‘fundamental rights (to know about one’s own genetic constitution and to make informed decisions about medical care)’. ¹³ Although there is no recognition of a right not to know in domestic law (Laurie, 2014b: 59), it has been accepted that doctors are not obligated to provide patients with information that they do not desire to know. ¹⁴ A right not to know genetic information pertaining to oneself is protected in two international instruments; however, neither is legally binding in the UK. ¹⁵ Whilst conceptualising the right not to know as being based on an individual’s interest in spatial privacy might also mean that it could be encompassed by Article 8 (Laurie, 2014a: 42), in the limited circumstances in which doctors can withhold information from their patients within this jurisdiction, this has not been legitimated on the basis of any such right. Rather, it is the doctor’s therapeutic privilege that has been cited, a privilege that entitles the doctor not to disclose certain information because they consider this to be objectively in the patient’s best interests, because disclosure would be detrimental to the patient’s health. ¹⁶ It thus appears that the law indirectly recognises an interest in not knowing rather than a right, albeit in very limited circumstances (Jackson, 2016: 212). For reasons that we will now begin to tease out, we would argue that, with one exception, interests offer a more constructive framing for the AI context.

The first note of caution to be expressed when considering whether a rights-based conceptualisation is appropriate for our particular disclosure dilemma is that a rights discourse may have a tendency to make conflicts seem more intractable. As both are prima facie entitlements of the same level, it might be assumed that neither the AI victim’s right to know nor her right not to know should trump the other, creating an impasse. However, it would be inaccurate to perceive rights as being necessarily absolute, or to contend that conflicts between rights cannot be resolved. Jurisprudence on settling conflict between Article 8 and 10 (privacy versus freedom of expression) under the ECHR, ¹⁷ for example, illustrates that legal solutions involving the balancing – or redefining (Scanlon, 2009: 76–78) – of rights can be found. Indeed, if it is accepted that conduct involving a decision to disclose or not disclose information to an individual engages Article 8, Article 8(2) offers a means of finding a balance between conflicting rights. It requires a pressing need, justification of the necessity of the interference and proportionate conduct. A rights-based conceptualisation could then offer a means of resolving the dilemma through an assessment of these factors, suggesting that an objection to a rights-based conceptualisation grounded on the intractability of rights can be contested.

When it comes to balancing the AI victim’s right to know and her right not to know, however, a particular difficulty is that the rights in question would be possessed by the one victim, rather than this being a case of balancing the conflicting rights of two or more individuals. ¹⁸ It is hard to envisage how, practically, the rights could be balanced when exercising one effectively cancels out the other. Moreover, as already noted, rights impose duties (Halpin, 1997: 30 and 261; Scanlon, 2009: 69–70 and 75). This problematises the AI disclosure dilemma further, because we must then conceive not only of victims’ rights to know and not to know, but also of duties to inform and not to inform that would obligate LEAs. The existence of these additional conflicting duties would only add further complications.

A further reason for a cautious assessment of the suitability of a rights discourse relates to this discourse’s dependency on a landscape that is amenable to it, such as is apparent within healthcare law. Is a growing recognition of rights also apparent in criminal justice? We can look to the push for greater victim involvement in the criminal justice process (Dignan, 2005: chapter 3), the greatest effect of which is seen in the paramountcy of victims’ rights in the United States since the emergence of the victims’ rights movement in the 1970s (Ginsberg, 2014), and calls from academics based in the UK from the 1980s onwards for turning away from victims’ needs to a rights-based approach (Mawby and Walklate, 1995; Mawby and Gill, 1987). We note also the significance of human rights for radical victimology (Kauzlarich et al., 2001). However, more specifically, although current policy in the UK recognises ‘rights’ (Walklate, 2007: 108) related to the bestowal of information under the Code of Practice for Victims of Crime (Ministry of Justice, 2015), these rights only apply after the victim has reported a crime. And they pertain to the progress of that particular crime and offender through the criminal justice system (see, e.g. Newton, 2003 and European Parliament and Council of the European Union (2012): Articles 4 and 6). These rights, and academic calls for victims’ rights including a right to knowledge (Mawby and Gill, 1987: 229), do not relate to a context in which LEAs have information regarding a crime of which a victim remains unaware.

Although there may be a case for advocating disclosure rights for unknowing crime victims, in the current absence of such a call, conceiving of such rights (and the accompanying conflicting duties that they could place on LEAs) seems premature. That said, we adopt a different position where the ‘knowing’ V1 has expressed a prior wish to know or not to know regarding future disclosures and this wish is ignored, because it is in her case that a rights-based claim could most persuasively be made. ¹⁹ As we argue in the next section, this exercise of her autonomy regarding private information about herself may well bring Article 8 rights under the ECHR into play, meaning that a right to know or not to know is offered legal recognition and protection. Moreover, there is no conflict between the right to know and not to know for her, because she has exercised her autonomy to decide which of these rights should prevail.

We suggest that the existence of rights to know and not to know is less likely to be accepted for unknowing V1–4 given the absence of any exercise of autonomy regarding future disclosures that could give rise to a duty on the part of LEAs to resolve the conflict between these rights one way or the other. Therefore, it would be more appropriate to conceptualise the issue as one involving unknowing victims’ interests. We recognise that utilising a different construct would not remove the conflict identified earlier; providing the unknowing AI victim with information to serve her interest in knowing would still frustrate her interest in not knowing. However, an interest-based framework would remove the potential added complication of creating conflicting duties on the part of LEAs. A resolution to the conflict between the interests could be achieved by achieving a balance between them. The starting point would be to recognise that they have equal significance given their connection to the same value enshrined in Article 8 of the ECHR of respecting private and family life. How they are to be balanced depends on the particular circumstances and issues at stake, so that one interest might outweigh the other in certain circumstances. And even though we have rejected a rights-based conceptualisation for unknowing victims, the criteria under Article 8(2) can still assist in this endeavour. It is thus necessary to consider the rationales for and against knowing, and their respective weights, so that these can then be situated within the facts of the particular case.

Should LEAs step into the AI victim’s private sphere?

Returning to Laurie’s conceptualisation of privacy as the foundation for the interests at stake in the disclosure dilemma, entering another’s state of privacy can sometimes be justified, but it can also constitute ‘unwarranted trespass’ (Laurie, 2014b: 59). Which of these two categories disclosing the information about the crimes committed against the AI victim falls into depends on the reasons for and against stepping into her private sphere (such as whether the disclosure is deemed necessary and whether there is a pressing need for it), and the consequences of disclosure. These are matters that we will now identify and reflect upon.

Conclusions: To know or not to know? The way forward and the implications for victimology

Our concern with unknowing AI victims in this paper has involved traversing unchartered territory for victimology. Exploring the parallels and disconnects with the debate regarding knowing and not knowing genetic information has revealed that conceptualising the disclosure dilemma as one involving rights to know and not to know is only appropriate for the knowing AI V1 who has expressed a prior wish regarding the issue. In her case, failing to respect her autonomy could well engage her right under Article 8 of the ECHR and it is highly unlikely that interference with this right could be justified on the basis of necessity or any pressing need. In all other cases involving unaware victims, the absence of any exercise of autonomy regarding future disclosures that could impose a duty upon LEAs to resolve the disclosure dilemma one way or the other, means there is an intractable conflict between purported rights to know and not to know. What is more, unlike the knowing V1 – who has effectively consented to and accepted the risk and burden of potential harm that knowledge brings with it – we would be imposing a right on an unknowing victim that might empower her but could (and in all likelihood will) cause her to suffer harmful effects. And we would be placing a serious burden upon LEAs to respect a right that we do not know the unknowing victim would want. This leads to the conclusion that the matter should be framed as one involving unknowing victims’ interests in knowing and not knowing, which we have argued are grounded primarily in spatial privacy. AI victims’ spatial privacy would always be violated by unsolicited disclosures because revealing the existence of the information affects their control of personal information, taking away their facility not to accept information about themselves. In contrast, the existence of autonomy-based foundations for these interests is more contentious, because unknowing victims cannot exercise any meaningful choice regarding information of which they are unaware.

Deciding whether the interest in knowing or not knowing should prevail is a challenging task. Herring and Foster (2012: 26) have contended that in disclosure dilemmas where we do not know whether the individual would wish to know the information in question, the starting point should be one of not informing. This mirrors Laurie’s (2004: 440) view that ‘[t]he presumption [should be] that individuals’ psychological privacy should be respected unless there is good reason not to do so’. As in the genetic context, there can be positive (albeit different) reasons for knowing the information in question, but in the AI context, disclosing the information is likely to make the unaware victim worse off overall. Arguments against disclosing on the basis of the invasion of spatial privacy and the high risk of consequent psychological harm that this is likely to cause victims outweigh arguments in favour of disclosure, such as empowering victims to seek reparation from offenders. In short, the burden of the disclosed information is greater than any potential empowerment it may provide. We are thereby advocating a form of modified paternalism regarding any now-adult victims of historic child sexual abuse and AI, given the probable, seriously harmful effects of an unsolicited disclosure. The exception to such an approach occurs in the case of the knowing V1 who has expressed a prior wish to be informed of future offences, because it is her autonomous choice to accept the risk of harm that knowing brings. That said, we noted in the introduction that LEAs are unlikely to be able to ascertain which victim (1–4) they are faced with, unless she has previously been identified and they are aware of the degree of her knowledge. Given the rationale behind the precautionary approach that we are advocating, we suggest that LEAs work on the assumption that the victim in question does not know that the AI exist and have been distributed unless and until this assumption is disproved.

To avoid disclosure dilemmas in the future where victims are aware of images and their distribution, it should be ensured that LEAs provide them with the opportunity to make a ‘timely and effectively informed’ (Laurie, 2014b: 54) decision to be told or not to be told about any subsequent offences involving the AI with the support of a counsellor. A further way of reducing future disclosure dilemmas in the case of known victims of child sexual abuse is for the possible existence of AI to be discussed sensitively with them during counselling and therapy sessions. Such discussions could reveal whether victims are aware of AI. Where no AI are known to exist, victims could be asked whether they would wish to be informed in the future if such images do come to light, and then supported through the process of making a meaningful decision on this. More problematic cases would be those in which counsellors are aware that AI exist but do not know whether the victim is cognisant of this, returning us to the same disclosure dilemma that has been the focus of this paper.

We have offered an ethical exploration of the issues that require consideration and suggested a way forward so that police policy can offer some means to resolve the conflict between AI victims’ interests in knowing and in not knowing. However, to provide a fully rounded answer to the immensely difficult question of whether it is appropriate to disclose, and the best way to do so, there is a need for further empirical research exploring practitioners’ and victims’ views to take forward this theoretical analysis. These voices offer a vitally important and necessary means of testing our conclusion that an approach based on modified paternalism is the most appropriate way forward in the case of unaware victims. Moreover, the involvement of victims would reduce the more paternalistic elements of our approach in this paper. However, in highlighting the need for such research, we must be wary of ‘trauma creep’ and of making generalisations regarding what victims’ autonomous views would be: it should not be assumed from the findings of any future study involving the experiences of some AI victims that the actual lived experiences and views of the collective of AI victims will be the same (see generally Walklate, 2016: 11).

Beyond our specific context, the analysis has broader implications for victimology. It has opened the door for the introduction of a new paradigm of the unknowing crime victim into victimology. We invite others to develop this victim typology and address questions such as: how should the parameters of this victim classification be defined? How would unknowing victims’ interests and needs be best met? Are members of certain groups in society more susceptible to becoming or being unknowing victims (see generally Walklate, 2007: 33–56)? In particular, should a right to know or not to know information about a crime committed against them exist for unknowing victims? Although we have focused on Article 8 of the ECHR, a more radical move would be to offer a guaranteed right to be notified for unaware victims under the Code of Practice for Victims of Crime and to bolster this right, and the other rights laid out in this Code, with the force of a statutory law as in the United States. Our analysis suggests that such an approach should be considered with caution because of the duty it would impose upon LEAs to disclose; is it safe to assume that unknowing crime victims would always wish to know of a crime committed against them? A right not to know might also be protected under the same statutory law, but presumably this could only come into play if victims had expressed a wish not to know in advance. Where such a wish has not been expressed, we have demonstrated that there may well be an interest in not knowing that would be defeated as soon as the duty to inform imposed by a right to know is acted upon.

Consideration should also be paid to why we might seek to disclose to other unaware victims that a crime has been committed against them; what purposes would disclosure serve? Factors against and in favour of disclosure will, of course, differ depending on the crime. For instance, the nature of the crime could have serious implications for a victim who was raped while unconscious (including the possibility that she may have contracted a sexually transmitted disease), making the case for disclosure strong, notwithstanding the undoubtable psychological harm that invading her spatial privacy will cause. The rape victim could also claim compensation from the Criminal Injuries Compensation Scheme, or possibly from the offender in a civil case. In contrast, consider the implications for an unaware victim of voyeurism who features in a video that the offender has created that has not been distributed and has been destroyed. Informing them of the crime committed against them would be likely to cause at least some level of mental distress and it is difficult to envisage what interest disclosure would serve. Would their understanding of their life experience be inadequate without this information? We very much doubt it. And there would be little likelihood of any form of compensation. The arguments in favour of non-disclosure and protecting the victim’s spatial privacy would appear much stronger in this case.

Finally, would the existence of this new paradigm challenge more conventional conceptions based on deeply rooted assumptions (see generally Walklate, 1989: xiv; 2007: 28) that a person has to suffer harm to be considered a victim of crime? ²³ This is especially pertinent, and challenging, given recent work demonstrating the connection between conceptions of the crime victim and trauma (McGarry and Walklate, 2015). Although seemingly at odds with contemporary criminal justice policy permeated by trauma narratives (McGarry and Walklate, 2015), we suggest it is necessary to consider whether there is a need for a broader conception of a victim encompassing an individual who has not yet been harmed and thus does not realise they have been victimised. Nonetheless, they have been wronged (Feinberg, 1984: 34–35) by a violation of their rights or interests and are highly likely to suffer harm if their victim status is revealed to them.