Skip to content
Search resources Open menu

Online Offense

Digital Evidence and What It Can Show About the User

Published

Snap Shot

  • The question: What can artifacts recovered from a device show about what the user knew, sought, and was sexually interested in?
  • What the research shows: User-saved and organized files, CSAM-specific search terms, repeated viewing, and chat logs are among the most informative evidence of deliberate acquisition. Within samples of people convicted of CSAM offenses, a greater proportion of images of boys is associated with sexual recidivism and appears in the CPORT.
  • What it does not show: A cached or thumbnail image, standing alone, shows only that it was displayed on the device; a recovered deleted file does not by itself show when it arrived, who placed it, or whether the user could access it. Collection features are group-level correlates; some are items in published risk tools (for example, the CPORT and the CASIC), and none alone establishes intent, a past contact offense, or dangerousness.
  • The law: Knowing possession or viewing can be proved circumstantially (Tucker, Romm, King), but files found only in caches or unallocated space, without evidence of knowledge or control, have led to reversals (Kuchinski, Flyer, Moreland). Examiner testimony falls under Rule 702.
  • Issues for each side: Prosecutors should charge files with the strongest evidence of user selection and not move from collection features to dangerousness without a clinical evaluation; defense counsel should obtain an independent examination early and ground evaluations in what the defendant actually selected and viewed, recognizing that the examination may confirm the government’s findings.

In cases involving child sexual abuse material (CSAM), the digital device is usually both the scene and the principal witness. Statutes and courts use the term “child pornography,” and this paper uses that term when describing the law; elsewhere it uses CSAM. A forensic examination can recover files that were saved, viewed, deleted, shared, or never opened, along with records of searches, communications, and the timing of activity. Those artifacts are routinely offered to prove what the defendant knew and intended, and they are increasingly used by psychologists and risk researchers to draw inferences about sexual interest and risk.

Both uses depend on a translation step. A file’s location, name, or timestamp is a technical fact; whether a particular person knew the file was there, sought it out, or was sexually interested in its content is a human fact. The translation is sometimes straightforward and sometimes not, and errors in either direction can affect verdicts and sentences.

This paper is written for attorneys and psychologists rather than forensic examiners. It describes the main categories of artifacts and what each can and cannot establish about the user, the case law on knowing possession when files are found in caches or deleted space, the research that uses collection features to assess risk, and the value of coordination between digital examiners and psychological evaluators.

Where Files Are Found and What Location Shows

Files saved by the user to folders such as documents or downloads, especially when renamed, moved, or organized, are the most direct evidence of deliberate acquisition and retention. Folder structures that sort material by age, sex, or content, and repeated access to particular files, indicate attention and interest by whoever created them.

Other locations require more interpretation. Web browsers automatically store copies of images from pages a user visits in a cache, typically without the user’s knowledge, and thumbnail databases created by operating systems record small versions of images that were displayed in a folder view. The presence of an image in a cache or thumbnail store shows that it was displayed on the device at some point; standing alone, it does not show that the user saw it, sought it, or knew it was stored. Context can supply what the location does not: search terms, the pages visited, how long the user remained on them, and whether similar material appears elsewhere on the device.

Deleted files and data in unallocated space present the hardest questions. When a file is deleted, the operating system typically marks the space as available without immediately overwriting the data, and forensic software can often recover the file or fragments of it. Recovery shows that the file existed on the device at some earlier time, and the deletion itself may be relevant, either as consciousness of guilt or as an effort to dispose of material that was unwanted. It does not by itself show when the file arrived, who placed it there, or whether the user could access it at the charged time, since an ordinary user cannot view data in unallocated space without specialized tools.

Shared devices and accounts complicate attribution in all of these locations. Evidence that ties activity to a particular person, such as user profiles, passwords, the timing of activity relative to the person’s documented whereabouts, and contemporaneous use of personal accounts, is often as important as the files themselves.

Peer-to-Peer Software and Distribution

Peer-to-peer file-sharing programs present distinctive interpretive issues. Most programs place downloaded files in a shared folder by default, making them available to other users without further action, and some allow sharing to begin before a download completes. A device may therefore hold partial downloads, files that were queued but never opened, and files whose names promised one kind of content and delivered another. Investigators commonly identify these devices through law enforcement versions of the software that download files directly from a single source and log the transaction.

Several facts help separate deliberate from incidental activity: whether the user changed default settings, for example to disable or limit sharing; whether files were moved from the shared folder or renamed; whether the user selected files by search terms associated with CSAM; how many files were downloaded and over what period; and whether files were opened, viewed, or retained. Courts have held that knowingly making files available in a shared folder, with awareness that others can download them, constitutes distribution when a download occurs (United States v. Shaffer, 2007; United States v. Budziak, 2012). The 2016 amendment to the federal Sentencing Guidelines requires that a defendant have “knowingly engaged in distribution” for the two-level distribution enhancement, which makes evidence of the user’s understanding of the software directly relevant at sentencing. In Budziak, the Ninth Circuit also held that the defense was entitled to discovery about the government’s investigative software when it had made a showing that the software’s operation was material to the defense.

Searches, File Names, Timestamps, and Communications

Search terms are among the most informative artifacts because they reflect what the user was looking for rather than what arrived. Terms that are specific to CSAM, or that combine age-related words with sexual terms, support an inference of deliberate seeking; generic adult search terms that returned mixed results support a different inference. File names can indicate content, but they can also be misleading; a file’s name was written by whoever first named it, not necessarily by the user, and names in shared material often exaggerate or misdescribe content. Forensic tools and investigators also use hash values, digital fingerprints that match files to known CSAM in databases maintained by the National Center for Missing & Exploited Children and law enforcement, which can identify content without viewing it but say nothing about how the file reached the device.

Timestamps record when files were created, modified, accessed, or downloaded, and they can establish patterns of activity over time. They must be interpreted carefully, because different file systems and programs record times differently, copying and syncing can change them, and time zones and device clocks may be inaccurate. Viewing duration, recorded by some programs and inferred from logs in others, can indicate sustained attention, though it does not distinguish attention from inactivity. Chat logs and messages are often the richest source of evidence about knowledge, intent, and conduct, including trading, discussions of sexual interest in children, and contact or attempted contact with minors; they can also show that the user expressed disinterest or that another person controlled the account.

The Law of Knowing Possession

The federal statutes require knowing possession, receipt, distribution, or access with intent to view, and the courts have applied that requirement to digital artifacts in a consistent way. In United States v. Tucker (2002), the Tenth Circuit held that a defendant who knew his browser cached images and intentionally sought out and viewed CSAM possessed the cached files. In United States v. Romm (2006), the Ninth Circuit held that a defendant who viewed and enlarged images, and who knew about and tried to delete his cache, exercised dominion and control over the cached files. By contrast, in United States v. Kuchinski (2006), the Ninth Circuit held that images in a cache could not be counted for sentencing when there was no evidence the defendant knew of the cache or could access or control it, although the 110 images he knowingly downloaded were properly counted. In United States v. Flyer (2011), the Ninth Circuit reversed a possession conviction based on images in unallocated space because there was no evidence that the defendant knew the files were there or could recover them at the charged time. In United States v. Moreland (2011), the Fifth Circuit reversed a conviction where several family members used the computers and the images were in unallocated space, holding that joint occupancy alone did not establish knowing possession.

Military law follows similar principles. In United States v. Navrestad (C.A.A.F. 2008), the court held that viewing images on a public computer without dominion or control is not possession. In United States v. King (C.A.A.F. 2019), it held that viewing does not require dominion or control and that circumstantial evidence, including password-protected devices, indicative search terms, the volume of related material, and admissions, could prove viewing even though the charged files were found only in unallocated space and browser cache. The Manual for Courts-Martial explanation of the child pornography offense similarly allows knowledge to be inferred from circumstantial evidence such as file names, search terms, and the number of images. The 2008 amendment adding “knowingly accesses with intent to view” to 18 U.S.C. §§ 2252 and 2252A reduced the government’s dependence on possession theories for online viewing, but it did not eliminate the requirement of knowledge and intent.

Digital Evidence in Risk Research

Researchers have proposed that digital evidence can inform psychological evaluations as well as prosecutions. Glasgow (2010) argued that forensic examinations can provide information relevant to risk assessment that is unavailable from self-report, such as search terms, the timing and duration of internet use for legal and illegal pornography, and chat logs that may distinguish fantasy-oriented from contact-oriented communication. Seto (2013) noted that court records typically describe only the images needed to prove the offense, so evaluators who rely on charging documents may miss information about the broader collection and the person’s other pornography use.

Several collection features have research support. In samples referred for clinical assessment, a CSAM charge has been a strong group-level indicator of sexual interest in children (Seto et al., 2006); a 2026 study of 516 men found wide variation among men convicted of CSAM offenses, so the charge alone does not establish interest or a diagnosis in any individual (Meißner et al., 2026). Within samples of people convicted of CSAM offenses, a greater proportion of images of boys than of girls is associated with sexual recidivism and appears as an item in the Child Pornography Offender Risk Tool (Seto & Eke, 2015); it also appears among the behavioral correlates of admitted sexual interest in children (Seto & Eke, 2017). Long, Alison, and McManus (2013) found that men with both image and contact offenses differed from men with image offenses only in the number, proportion, and type of images they held, as well as in previous convictions and access to children. Fortin and Proulx (2019), analyzing more than 61,000 images and their metadata from 40 men convicted of possession, found four patterns of change in collecting over time; the most common was a progressive decrease in the age of children depicted and an increase in the severity of acts, but not all collections followed that path; this describes how some collections changed, not a pathway to contact offending, and the research does not support a universal sequence and does not rule out change in an individual. The risk tools that use these features, and their limits, are discussed in the Psychosexual Risk Evaluations hub.

Cautions About Inferring Sexual Interest from Collections

Collection features are group-level correlates, not diagnostic markers, and several cautions apply when they are used to draw conclusions about a particular person.

First, what is on a device is not necessarily what the user chose. Bulk downloads, archives containing thousands of files, peer-to-peer results that did not match their file names, and automatically cached material can place content on a device that the user never viewed. Collection features computed from everything recovered may therefore reflect the behavior of the software or of other people as much as the interests of the user. Evidence of selection, such as searches, organization, repeated viewing, and retention, is more informative than counts.

Second, the proportions that have research support were measured in particular samples using particular coding rules, and many studies coded the material that police happened to classify. Collections in current cases are often far larger and more varied, and coders do not always agree about age or content (Kloess et al., 2019). Third, the absence of a feature does not establish the absence of interest, and the presence of a feature does not establish a diagnosis; diagnosis requires a clinical evaluation that considers all available information. Fourth, no collection feature alone establishes whether a person has committed or will commit a contact offense, which is the question of greatest concern to courts.

Coordination Between Digital and Psychological Experts

Because the meaning of digital evidence depends on technical details that psychologists are not trained to evaluate, and the significance of those details for sexual interest and risk depends on clinical and research knowledge that digital examiners are not trained to apply, the two disciplines work best together. A digital examiner, whether retained by the defense, employed by the government, or appointed by the court, can determine where files were found, whether they were opened, how they arrived, what searches preceded them, how the software was configured, and who was using the device at the relevant times. A psychological evaluator can then interpret that information in light of the research on sexual interest and risk, compare it with the person’s self-report, and identify discrepancies that warrant further inquiry.

In practice, coordination means asking the digital examiner specific questions: which files were user-selected rather than automatically stored, what proportion of the content was ever opened, what search terms were used, whether there is evidence of trading or communication with minors, and how activity was distributed over time. It also means that the psychological evaluator should not compute collection features from summary charts or charging documents without knowing how the underlying data were produced. Under 18 U.S.C. § 3509(m), CSAM remains in government custody and is made available to the defense for inspection at a government facility, which affects the time and resources needed for an independent examination and should be considered early in a case.

Limits of the Evidence

Digital forensic findings are only as reliable as the tools and methods used to produce them, and both change rapidly. Different forensic programs can parse the same data differently, timestamps and artifacts vary across operating systems and applications, and encryption, cloud storage, and automatic deletion increasingly limit what can be recovered. The research linking collection features to sexual interest and recidivism is based on samples of identified offenders from particular jurisdictions and periods, uses official records that undercount reoffending, and reports modest effect sizes.

Neither a forensic report nor a risk study can establish what a particular person knew, intended, or will do. Digital artifacts are circumstantial evidence; their significance depends on how well they are tied to the person, how they fit with other evidence, and whether alternative explanations have been examined. A psychological opinion that rests on collection features should state how those features were derived and what uncertainty attends them.

Questions to Ask About This Research

When a greater proportion of images of boys is offered as a sign of sexual interest or risk (Seto & Eke, 2015), ask what exactly was counted, because a feature computed from everything recovered, including bulk downloads and cached files, may not reflect what the user selected, and ask whether a group-level association applies to this person. When Fortin and Proulx (2019) are cited to show that a decreasing age of children in a collection marks escalation, ask how much evidence the finding rests on, since it drew on 40 men and not all collections followed that path. When the absence of a feature, or the fact that coders do not always agree (Kloess et al., 2019), is offered to discount the content of a collection, ask whether the conclusion follows: the absence of a feature does not establish the absence of interest, and disagreement among coders calls for checking how the disputed images were coded rather than assuming every classification is uncertain. These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.

Digital Evidence and the Law

Federal and Military Standards

Testimony by digital forensic examiners is expert testimony under Federal Rule of Evidence 702, as amended December 1, 2023, which requires the proponent to show by a preponderance of the evidence that the expert’s opinion rests on sufficient facts or data and reflects a reliable application of reliable methods. Questions about which tool was used, whether its output was validated, and whether the examiner can explain how an artifact was created go to admissibility as well as weight under the amended rule. Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993) and Kumho Tire Co. v. Carmichael (1999) apply to technical as well as scientific testimony. Rule 704(b) bars an expert from stating whether the defendant had the knowledge or intent that the offense requires; a digital examiner may describe the artifacts and what they are consistent with, but whether the defendant knowingly possessed material is for the jury. Budziak recognizes that the defense may obtain discovery about investigative software when it makes a sufficient showing of materiality, and Ake v. Oklahoma (1985) recognized a due process right to psychiatric assistance in defined circumstances; it did not announce a general right to non-psychiatric experts such as digital-forensic examiners, and courts differ on extending it. In courts-martial, Military Rule of Evidence 702 parallels the federal rule; Military Rule of Evidence 704 has no counterpart to Federal Rule 704(b), although relevance, helpfulness, and Rule 403 still limit mental-state opinions, Military Rule of Evidence 1102 generally incorporates amendments to the Federal Rules after 18 months, and an accused who shows necessity may obtain expert assistance under Rule for Courts-Martial 703 and United States v. Gonzalez (C.M.A. 1994). The December 1, 2023 amendment to Federal Rule 702 accordingly became part of Military Rule of Evidence 702 in 2025: the proponent of expert testimony in a court-martial must now show the military judge that it is more likely than not that each admissibility requirement is met, so gaps in an expert’s basis or in the application of the method are questions of admissibility rather than of weight for the members (Joint Service Committee on Military Justice, 2025).

Arizona

Arizona prosecutions under A.R.S. § 13-3553 require proof that the defendant knowingly possessed, received, distributed, or otherwise dealt in the depiction, and the same questions about caches, deleted files, shared devices, and file-sharing arise. Arizona Rule of Evidence 702, amended effective January 1, 2012, follows the federal rule and Daubert, and Arizona Rule 704(a) permits ultimate-issue opinions, but Rule 704(b) bars an expert in a criminal case from opining that the defendant did or did not have the required mental state; further limits on mental-state testimony come from State v. Mott (1997). Because each image may be charged as a separate count and dangerous crimes against children sentences run consecutively under § 13-705, forensic questions about which specific files were knowingly possessed can determine not only guilt but the length of the sentence.

Special Issues in Legal Practice

The principles described above apply with equal force to every party in a legal proceeding. The science operates the same way regardless of who is asking the questions, and the research summarized here does not by itself favor either side. What differs between the parties is not the science but the vantage point — the particular places where these issues tend to surface, and the particular errors each party is best positioned to guard against. The two summaries below identify those recurring issues. They are offered for orientation, not as litigation strategy; in every case the controlling question is what the evidence in that case will, and will not, support, and the expert’s role is to inform the trier of fact rather than to advocate for either side.

Special Issues for Prosecutors

Digital evidence can provide strong proof of knowledge and intent: user-saved and organized files, CSAM-specific search terms, repeated viewing, changes to file-sharing settings, chat logs showing trading or discussion of sexual interest in children, and evidence tying activity to the defendant’s accounts and whereabouts. Tucker, Romm, and King show that such circumstantial evidence can establish possession or viewing even when files are found in caches or unallocated space. The research also supports treating some collection features, such as a predominance of images of boys or, in some collections, a progression toward younger children, as relevant to sexual interest. Prosecutors are often best positioned to ensure that charged files are those with the strongest evidence of user selection and that forensic findings are explained in terms the jury can evaluate.

The research and case law also counsel caution. Files found only in caches, thumbnail stores, or unallocated space, without evidence of knowledge or control, have led to reversals in Kuchinski, Flyer, and Moreland. Counts that include partial or unopened downloads, or files on shared devices without attribution evidence, are vulnerable. Expert testimony that moves from collection features to conclusions about the defendant’s sexual interest or dangerousness, without a clinical evaluation and without accounting for how the collection was assembled, may overstate what the research supports.

Special Issues for Defense

The research and case law support careful examination of how each charged file reached the device and whether the defendant knew of it. Automatic caching, default peer-to-peer sharing, bulk downloads, misleading file names, shared devices, and deleted files that could not be accessed at the charged time are recognized grounds for questioning knowledge, as Kuchinski, Flyer, and Moreland illustrate, and the knowledge requirement for the distribution enhancement makes the defendant’s understanding of file-sharing software relevant at sentencing. Defense counsel are often best positioned to obtain an independent forensic examination early, to seek discovery about investigative tools where Budziak applies, and to ensure that any psychological evaluation rests on accurate information about which material the defendant actually selected and viewed.

The research also counsels caution about the limits of these arguments. Circumstantial evidence such as search terms, organization, and repeated viewing can establish knowledge even when particular files are in caches, as Tucker and Romm show, or in unallocated space, as King shows, and the “access with intent to view” provisions reach online viewing without download. An independent examination may confirm rather than undermine the government’s findings, and it may reveal additional material or communications. Collection features associated with sexual interest in children are items in published risk tools (for example, the CPORT and the CASIC), and an evaluation that disregards them without explanation may be incomplete.

This paper is for educational purposes only and is not legal advice. The author is a psychologist, not an attorney; summaries of law are general and must be verified against current authority or with a licensed attorney.

References

Fortin, F., & Proulx, J. (2019). Sexual interests of child sexual exploitation material (CSEM) consumers: Four patterns of severity over time. International Journal of Offender Therapy and Comparative Criminology, 63(1), 55–76. https://doi.org/10.1177/0306624X18794135

Glasgow, D. (2010). The potential of digital evidence to contribute to risk assessment of internet offenders. Journal of Sexual Aggression, 16(1), 87–106. https://doi.org/10.1080/13552600903428839

Joint Service Committee on Military Justice. (2025). Amendments to the Military Rules of Evidence since 1984 (G. E. Maggs, Ed.; 2025 ed.). https://jsc.defense.gov/Portals/99/MRE%20Amendments%202025.pdf

Kloess, J. A., Woodhams, J., Whittle, H., Grant, T., & Hamilton-Giachritsis, C. E. (2019). The challenges of identifying and classifying child sexual abuse material. Sexual Abuse, 31(2), 173–196. https://doi.org/10.1177/1079063217724768

Long, M. L., Alison, L. A., & McManus, M. A. (2013). Child pornography and likelihood of contact abuse: A comparison between contact child sexual offenders and noncontact offenders. Sexual Abuse, 25(4), 370–395. https://doi.org/10.1177/1079063212464398

Meißner, L., Domany, S., Rathgeber, F., Schmidt, A. F., Banse, R., & Eher, R. (2026). Sexual interests of men with different child sexual offenses as assessed by direct and indirect measures: Do users of child sexual abuse material differ? Sexual Abuse, 38(6), 677–706. https://doi.org/10.1177/10790632261454316

Seto, M. C. (2013). Internet sex offenders. American Psychological Association. https://doi.org/10.1037/14191-000

Seto, M. C., Cantor, J. M., & Blanchard, R. (2006). Child pornography offenses are a valid diagnostic indicator of pedophilia. Journal of Abnormal Psychology, 115(3), 610–615. https://doi.org/10.1037/0021-843X.115.3.610

Seto, M. C., & Eke, A. W. (2015). Predicting recidivism among adult male child pornography offenders: Development of the Child Pornography Offender Risk Tool (CPORT). Law and Human Behavior, 39(4), 416–429. https://doi.org/10.1037/lhb0000128

Seto, M. C., & Eke, A. W. (2017). Correlates of admitted sexual interest in children among individuals convicted of child pornography offenses. Law and Human Behavior, 41(3), 305–313. https://doi.org/10.1037/lhb0000240

Legal Authorities

Ake v. Oklahoma, 470 U.S. 68 (1985).

Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).

Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).

United States v. Budziak, 697 F.3d 1105 (9th Cir. 2012).

United States v. Flyer, 633 F.3d 911 (9th Cir. 2011).

United States v. Kuchinski, 469 F.3d 853 (9th Cir. 2006).

United States v. Moreland, 665 F.3d 137 (5th Cir. 2011).

United States v. Romm, 455 F.3d 990 (9th Cir. 2006).

United States v. Shaffer, 472 F.3d 1219 (10th Cir. 2007).

United States v. Tucker, 305 F.3d 1193 (10th Cir. 2002).

United States v. Gonzalez, 39 M.J. 459 (C.M.A. 1994).

United States v. King, 78 M.J. 218 (C.A.A.F. 2019).

United States v. Navrestad, 66 M.J. 262 (C.A.A.F. 2008).

State v. Mott, 187 Ariz. 536, 931 P.2d 1046 (1997).

18 U.S.C. § 2252.

18 U.S.C. § 2252A.

18 U.S.C. § 3509(m).

U.S. Sentencing Guidelines Manual § 2G2.2(b)(3) (U.S. Sentencing Comm’n 2025).

10 U.S.C. § 934 (Article 134, Uniform Code of Military Justice).

Manual for Courts-Martial, United States, pt. IV, ¶ 95 (2024 ed.).

R.C.M. 703.

Ariz. Rev. Stat. Ann. § 13-705.

Ariz. Rev. Stat. Ann. § 13-3553.

Fed. R. Evid. 702.

Fed. R. Evid. 704(b).

Ariz. R. Evid. 702.

Ariz. R. Evid. 704.

Mil. R. Evid. 702.

Mil. R. Evid. 704.

Mil. R. Evid. 1102.

Related topic papers: Child Sexual Abuse Material: Federal, Arizona, and Military Law · Culpability and Sentencing Factors in CSAM Cases · Estimating Age from Images · Risk Assessment Tools for Online Offenders · Expert Testimony in Online Sexual Offense Cases

Related topics from Arizona Forensic Services: Psychosexual Risk Evaluations · Autism Online Offense · Courts-Martial