Developing Software to Estimate Age in Child Pornography Images for Forensic Purposes: Relevance and Limitations in Psychocriminology

Sir, Child pornography is mainly disseminated by the Internet through pedophile networks, notably peer-to-peer file exchanges (1). The arrest of people who participate actively (production and dissemination of images) or passively (downloading and possessing images) in child pornography is hindered by the international aspect of this offense. Cooperation between countries is thus needed and has already led to the arrest of numerous child pornographers (2,3). However, countries do not all have the same legal measures to deal with child pornography material (2,4). Almost all states forbid pornographic images involving minors (4–6). In 2008, Interpol found 680 victims in 520,000 images in its database (7). Very often, these images are the only proof of the offense. Child pornographers are generally identified when they disseminate and download images. They then try to justify themselves by saying that the characters depicted are adults. It is thus essential in these situations to prove that they are in fact minors. There are a number of articles in the literature which describe the difficulty of proving minority, notably using Tanner’s criteria (8–12). Other techniques for estimating age have thus been developed (13–16). This will help to assess whether erotic or pornographic images involve minors (17,18). These computer-based methods of estimating age raise questions about their relevance and limits in psychocriminology. Three specific types of child pornography images need to be considered: (i) those that use minors, (ii) those that use virtual images and pseudo-photographs, and (iii) those that use adults.