United States Court of Appeals For the Eighth Circuit ___________________________

No. 23-3504 ___________________________

United States of America

lllllllllllllllllllllPlaintiff - Appellee

v.

Clint Robert Schram

lllllllllllllllllllllDefendant - Appellant ____________

Appeal from United States District Court for the Western District of Missouri - St. Joseph ____________

Submitted: January 14, 2025 Filed: February 12, 2025 ____________

Before LOKEN, ARNOLD, and KELLY, Circuit Judges. ____________

ARNOLD, Circuit Judge.

After a jury convicted Clint Schram of multiple offenses stemming from his operation of child pornography websites, the district court1 sentenced him to a life term and four concurrent thirty-year terms of imprisonment. He contends that the evidence was insufficient to support his convictions, that the district court

1 The Honorable Stephen R. Bough, United States District Judge for the Western District of Missouri. improperly admitted multiple images of child pornography into evidence, and that the district court erred in calculating his guidelines sentencing range and imposing an overlong sentence. Because we discern no reversible error, we affirm.

In 2020, a grand jury returned an indictment, which, in relevant part, charged Schram with four counts of advertising child pornography and one count of engaging in a child exploitation enterprise. Advertisement is shorthand for knowingly making, printing, publishing, or causing to be made, printed, or published a “notice or advertisement seeking or offering” to “receive, exchange, buy, produce, display, distribute, or reproduce, any visual depiction, if the production of such visual depiction involves the use of a minor engaging in sexually explicit conduct and such visual depiction is of such conduct.” 18 U.S.C. § 2251(d)(1). As charged, a child exploitation enterprise, in turn, consists of a series of advertising offenses committed “in concert with three or more other persons” that constitute “three or more separate incidents” and involve “more than one victim.” Id. § 2252A(g)(2).

Schram stood trial, and the government presented evidence that he administered four sites where he and other users shared links to what appeared to be child pornography. The jury then found Schram guilty on the five counts at issue.

Schram insists that the evidence was insufficient to convict him, but we are unconvinced. The gap in the evidence, according to Schram, was proof that he advertised depictions of real children engaged in sexually explicit conduct. The parties do not dispute that advertising depictions of real children, rather than computer-generated facsimiles, was essential to Schram’s convictions. But we agree with the government that the record was against Schram.

It is enough to uphold the convictions if any reasonable jury could have found Schram guilty beyond a reasonable doubt. United States v. Koch, 625 F.3d

-2- 470, 478 (8th Cir. 2010). That is a matter we review de novo, id., notwithstanding some suggestion that Schram failed to dispute the sufficiency of the evidence in the district court. If the suggestion were true, we would review only for plain error, see United States v. Clarke, 564 F.3d 949, 954 (8th Cir. 2009), but there is no need to pursue the point since the evidence was sufficient regardless of how we look at it.

That is so because the jury saw excerpts from pornographic content Schram advertised on his websites, all of which featured children. It could, therefore, inspect those excerpts and decide for itself that the children were real. We have said as much before, and so have our sister circuits. Koch, 625 F.3d at 479; see, e.g., United States v. Pawlak, 935 F.3d 337, 350 (5th Cir. 2019); United States v. Sims, 428 F.3d 945, 957 (10th Cir. 2005); cf. also United States v. Vig, 167 F.3d 443, 449 (8th Cir. 1999).

With improvements in image-generation technology, we may someday have to revisit our precedent, cf. Ashcroft v. Free Speech Coal., 535 U.S. 234, 259 (2002) (Thomas, J., concurring in the judgment), but Schram has not convinced us that today is that day. We will suppose, for present purposes, that we could disregard our court’s prior decisions if changes in technology undermined their assumption that jurors can reliably distinguish images of real children from images of virtual children. But see United States v. Rodriguez-Pacheco, 475 F.3d 434, 442 (1st Cir. 2007). The trouble for Schram is that the record here, far from undermining that assumption, is entirely consistent with it.

Until this appeal, that was obvious because the record was devoid of evidence about the distinguishability of real and virtual children. And it is only slightly less obvious now, after Schram pointed us to a handful of webpages about computer image generation. Those webpages, which we assume we can consider, are less than illuminating. They reveal, at most, that computer programs could generate images of virtual people at the time of Schram’s offense and that their

-3- realistic output was becoming ever more realistic. Few could dispute that, but it is beside the point. What matters is how accurate jurors are in distinguishing realistic images of virtual children from images of real children. See id. at 443 n.8. And on that question, the webpages are silent.

Without more, we are not prepared to depart from our court’s precedent allowing juries to decide whether images depict real children based on the images themselves. At the foundation of that precedent is the principle that the government need not produce evidence to negate a speculative assertion that a child in an image is virtual. Vig, 167 F.3d at 450. That principle is as true today as it was when we announced it—three years after Congress found that one could make images of virtual children almost “indistinguishable to the unsuspecting viewer” from images of actual children. Child Pornography Prevention Act, Pub. L. No. 104-208, § 121, 110 Stat. 3009, 3009-26 (1996). And it requires us to reject Schram’s challenge to the sufficiency of the evidence. On the nearly empty record here, Schram’s concern that images shown to the jury depicted virtual children is just speculation unsupported by any concrete facts.

Though our conclusion does not depend on it, we note as well that research on detecting images of virtual children leaves considerable doubt about the risk that the jury mistook an image of a virtual child for an image of a real child here. Even years after Schram’s offenses, less than one percent of child sexual abuse material was both computer generated and photorealistic. At least that was the percentage reported by the Stanford Internet Observatory and the nonprofit Thorn based on files Thorn sampled from “communities dedicated to child sexual abuse.” David Thiel et al., Generative ML and CSAM: Implications and Mitigations, at 2 (2023). And of that percentage, only about two-thirds of files were “highly photorealistic,” though even those could still be “visually distinguished as being generated.” Id. at 3. In our view, this is confirmation that the danger of confusing virtual children with real children at Schram’s trial was speculative, however serious it might be at some later date.

-4- We are also unpersuaded that we should reverse Schram’s convictions because the district court admitted multiple images of child pornography into the record. We review the admission of the images of child pornography for abuse of discretion, United States v. Becht, 267 F.3d 767, 770 (8th Cir. 2001), again declining to decide whether plain error review might apply since it would not change our conclusion. Schram contends that admitting the child pornography violated Federal Rule of Evidence 403, which provides in relevant part that a district court may exclude evidence if the danger that the evidence would be needlessly cumulative or unfairly prejudicial substantially outweighs its probative value. The district court, however, had the discretion to determine that the child pornography helped prove Schram’s advertisement offenses without unduly prejudicing Schram or duplicating the effect of other evidence.

As to some of the admitted child pornography, this was an easy determination to make. The government introduced a screenshot of one of Schram’s websites displaying an image of naked children and five exhibits containing images of child pornography from pictures or videos Schram linked on his websites. All these images were evidence that Schram used his websites to advertise child pornography, something the government had to prove to convict him. Schram protests that the government could have proven the same thing with just four images—one for each website on which he advertised—but nothing in Rule 403 forced the government to pick a single image from each website and discard the rest. Rule 403 does not require the government to produce only the minimum amount of evidence necessary to prove its case. What it requires is that the district court strike a balance between the probative value of the evidence that was offered and its tendency to encourage a verdict on improper grounds or make litigation inefficient. We have accordingly held that the government need not restrict its presentation of child pornography to one image for each device on which a defendant stores it. See United States v. Blanks, 985 F.3d 1070, 1074 (8th Cir. 2021); see also United States v. Naidoo, 995 F.3d 367, 376 (5th Cir. 2021) (per curiam). And here we similarly hold that the government’s presentation of

-5- Schram’s child pornography could feature more than one image from each of his websites. The images the government showed were few and highly probative, and the district court’s decision to admit them was reasonable.

Other images of child pornography that the government introduced were less probative, but we do not believe the district court committed reversible error in admitting them. The government showed the jury exhibits containing screenshots from videos found on Schram’s hard drive. The screenshots depicted girls exposing their vaginas or anuses, sometimes while digitally penetrating themselves. Unlike the images from Schram’s websites, only some of the screenshots were from videos linked on Schram’s websites, and the government does not contend that Schram himself obtained any of the videos from his websites or linked them there. The screenshots thus did not evidence any particular child pornography advertisement Schram made. But they did further the government’s case less directly, and we are confident that any mistake in admitting them did not influence Schram’s convictions.

The screenshots tended to prove Schram’s propensity to advertise child pornography, for the jury could find, as Schram’s trial counsel conceded, that Schram collected the screenshots because he was “a coveter and a keeper of child pornography.” And if it so found, the jury would have had greater reason to infer that Schram used his website to obtain or exchange child pornography, which would have qualified as prohibited advertising. 18 U.S.C. § 2251(d)(1)(A); see United States v. Heatherly, 985 F.3d 254, 269 (3d Cir. 2021).

Supporting this inference was a proper purpose for introducing the screenshots. Though we typically frown on offering evidence of a defendant’s uncharged misdeeds to show his propensity to commit charged misdeeds, see United States v. Furman, 867 F.3d 981, 988 (8th Cir. 2017), we do not do so here. Federal Rule of Evidence 414(a) explains why. It authorizes district courts to admit evidence that a criminal defendant accused of “child molestation”

-6- committed any other act of “child molestation,” and it permits consideration of the evidence “on any matter to which it is relevant.” Since child molestation includes both possessing child pornography and advertising it, Fed. R. Evid. 414(d)(2)(B), introducing screenshot evidence that Schram possessed child pornography as part of the proof that he advertised child pornography falls squarely within the rule.

We must still consider whether the screenshots were so unfairly prejudicial or unduly cumulative that their admission was improper under Rule 403, see United States v. Splettstoeszer, 956 F.3d 545, 547 (8th Cir. 2020), but we think that doubtful. As to unfair prejudice, it is true that images “depicting child pornography are by their very nature disturbing, and viewing such depictions is highly likely to generate an emotional response. But that alone cannot be the reason to exclude the evidence.” United States v. Evans, 802 F.3d 942, 946 (8th Cir. 2015). The screenshots were not especially graphic or upsetting. See id.; cf. Heatherly, 985 F.3d at 267–68. The jury had also seen worse images linked by Schram on his websites, including depictions of men having vaginal intercourse with girls and ejaculating in and on them. And the district court cautioned the jury not to convict Schram simply because he may have committed an uncharged child pornography crime. See United States v. Burch, 113 F.4th 815, 820 (8th Cir. 2024).

We are slightly more troubled by the redundancy of the screenshots with the images of child pornography from Schram’s websites that the jury also viewed. Yet we hesitate to say that the screenshots were too cumulative to admit. There were only nine exhibits containing screenshots, only a few screenshots in each exhibit, and only a few seconds for the jury to see each screenshot when published. See United States v. Worthey, 716 F.3d 1107, 1114–15 (8th Cir. 2013). Considering the limited presentation of screenshots and the more disturbing child pornography already in the record, any error in admitting the screenshots was, at the very least, harmless.

-7- Schram next challenges his sentence, though no more persuasively than he challenged his convictions. Without reaching its merits, we can first reject Schram’s argument that the district court improperly enhanced the offense level used to calculate his guidelines sentencing range. According to Schram, the district court erred by applying USSG § 3C1.1, which prescribes a two-level enhancement for willfully obstructing, impeding, or attempting to obstruct or impede “the administration of justice with respect to the investigation, prosecution, or sentencing” of an offense of conviction. We are more than a little skeptical of this argument since, before trial, Schram sent a letter to the magistrate judge assigned to his case in which he threatened to kill her if she did not dismiss the charges against him. See United States v. Eye, 520 F. App’x 852, 853 (11th Cir. 2013) (per curiam); cf. United States v. Wahlstrom, 588 F.3d 538, 544 (8th Cir. 2009). But in any event, Schram had earned the maximum offense level even before the district court applied the enhancement, so the enhancement caused him no harm. United States v. Jensen, 834 F.3d 895, 902 (8th Cir. 2016); United States v. Hamilton, 929 F.3d 943, 948 (8th Cir. 2019).

Just a little more discussion is necessary to reject Schram’s argument that his sentence was substantively unreasonable. The substantive reasonableness of a sentence is something we review for abuse of discretion, and, because Schram’s sentence was within the guidelines range, it was presumptively reasonable. United States v. Goodale, 738 F.3d 917, 925–26 (8th Cir. 2013). Nothing in the record rebuts that presumption; in fact, the record more than amply supports Schram’s sentence. Schram ran four websites used to share child pornography. His other behavior confirmed his sexual interest in children while also revealing his violent tendencies. In addition to his threats against the magistrate judge, there is evidence that Schram planned to kidnap, enslave, rape, and perhaps to kill an eight-year-old girl; threatened to kill a coworker, rape his children, and kill his family; and plotted to bomb a union hall. While incarcerated, Schram threatened to kill a prison guard, tried to strangle a cellmate, kicked a fellow inmate, and drew sexually explicit images of children on his cell walls and in a booklet. On one

-8- occasion, he asserted that he had raped and murdered multiple children in the past. Giving due weight to Schram’s lack of prior convictions and his mental health and substance abuse problems, the district court could readily determine that a lengthy sentence was necessary to reflect the seriousness of Schram’s offense and eliminate the danger he posed to the public. See 18 U.S.C. § 3553(a)(2)(A), (C); cf. United States v. Dehghani, 550 F.3d 716, 719, 723 (8th Cir. 2008); United States v. Cottrell, 853 F.3d 459, 463 (8th Cir. 2017).

Though Schram emphasizes it, the fact that users of his websites received shorter sentences than his does not undermine this conclusion. For one thing, when we consider sentencing disparities, we are, with an exception inapplicable here, concerned only with national disparities. United States v. Heard, 91 F.4th 1275, 1280 (8th Cir. 2024). But even if we set that point aside, “disparate sentences among dissimilar defendants are not unwarranted,” United States v. Driscoll, 122 F.4th 1067, 1071 (8th Cir. 2024), and Schram and the users were highly dissimilar. The users were at most staff members of Schram’s websites. Each pleaded guilty to a single count of advertising child pornography, and each received a sentence of about twenty years’ imprisonment. Schram, by contrast, ran all four websites, stood trial, and was convicted of four counts of advertising child pornography, plus another count of engaging in a child exploitation enterprise. Even disregarding Schram’s troubling uncharged conduct, we would expect his sentence to be much longer than the users’.

Since Schram has no further objections to his sentence, we conclude that the district court did not commit reversible error in imposing it. Schram’s convictions and sentence must stand.

Affirmed. ______________________________

-9-