Iowa Supreme Court Says Speculation That a Video Is a Deepfake Does Not Bar Admission

What did the Iowa Supreme Court decide?

It affirmed. The district court did not abuse its discretion in admitting a video of the abuse of a sleeping victim, even though, as the court put it, "there was no chain-of-custody evidence connecting the video to the defendant" (slip op. 2). The defendant was convicted of third-degree sexual abuse and received a ten-year indeterminate sentence (slip op. 5). He had argued that the video was not authenticated, was vulnerable to deepfake technology, and was barred by the best-evidence rule because it was a copy (slip op. 4).

Can a video be authenticated without a firsthand or silent witness?

Yes, on this decision. The defendant read State v. Manning, 26 N.W.3d 385 (Iowa 2025), as recognizing only those two routes. The court disagreed: "There is no special codicil to the rules of evidence governing videos" (slip op. 9). Circumstantial evidence under Iowa R. Evid. 5.901(b)(4) can suffice.

Here the victim testified to recognizing her bedroom, bed, clothing and the defendant's tattooed hand. She could not personally verify the abuse itself. The video surfaced about a month later, and the defendant blocked her when she contacted him about it (slip op. 10). The court asked, "What third-party fabricator would have access to all that?" (slip op. 10).

The court also recorded that no search warrant was obtained for the defendant's devices and that the investigating officer was unable to retrace the video's digital footprint (slip op. 4). The opinion does not describe any forensic analysis of the file and does not say whether one was done. That describes this record. The court did not say such analysis is unnecessary elsewhere.

Who had to raise the deepfake question under the duplicate rule?

The defendant. Rule 5.1003 makes a duplicate admissible "unless a genuine question is raised about the original's authenticity." The court agreed with the State that the defendant "didn't raise a genuine question, only conjecture" (slip op. 14), and that he "had the initial burden to raise a genuine question about the authenticity of the original" (slip op. 16).

Defense counsel's closing began, "Is it fake? Is it real? Ladies and gentlemen, we live in the age of artificial intelligence" (slip op. 14). The court said the advocacy "rested on popular culture rather than any of the evidence in the case" (slip op. 14). It also used the phrase "liar's dividend," which its footnote defines as the possibility that public awareness of deepfakes makes it easier to raise doubts about legitimate evidence (slip op. 15 n.4).

The court did not dismiss the risk: "We are not oblivious to the risks of video evidence" (slip op. 13). It cited the New York Court of Appeals decision In re M.S. (Feb. 17, 2026) as a contrast. As Iowa describes it, New York's highest court held that videos obtained by a self-described hacker and given to law enforcement well after the events were not sufficiently authenticated. We have not read In re M.S. The Iowa court's point was that here "there are no identifiable circumstances to suggest that the video might be a deepfake" (slip op. 13).

What did the court say about proposed Federal Rule of Evidence 901(c)?

Only what the State raised. The opinion says the Advisory Committee on Evidence Rules "has been considering" a possible new Rule 901(c), citing the committee's May 7, 2026 agenda book. The court describes the draft as requiring a party challenging evidence as AI-fabricated to present evidence sufficient to support a finding of fabrication, after which the proponent would have to show the item is "more likely than not authentic" (slip op. 16). The court added that Federal Rule 1003 and Iowa Rule 5.1003 "already work this way" in a sense (slip op. 16).

We did not open the agenda book and say nothing about the proposal's current text or status.

What does it mean for defense counsel, prosecutors and evidence practitioners?

This is our reading of one state decision. The holding rests on a record with a victim who identified private detail, a timeline tying the video to the defendant's presence, and a defense that offered no evidence of manipulation. A different record could come out differently, as the court's own contrast with In re M.S. shows.

Defense. On the court's reasoning, a general objection that AI exists did not carry the motion. The opinion does not say what evidence is enough, and it does not say an expert is required. Any list of record indicators, such as an unexplained source or a provenance gap, is our suggestion and is not drawn from the decision.

Prosecution. Authentication here did not depend on chain of custody or on a witness who saw the recording made. The court also noted that, as a practical matter, the State had to convince the jury beyond a reasonable doubt that the video was legitimate, because "the State's case depended on the video" (slip op. 16).

E-discovery and digital evidence. The opinion reports that no search warrant was obtained for the defendant's devices and that the officer was unable to retrace the video's digital footprint (slip op. 4). The video was still admitted on the testimony and circumstances in this record. What a proponent needs elsewhere depends on that jurisdiction's rules.

What should counsel do now?

• Iowa counsel: read slip op. 8 to 16 before briefing an authentication or best-evidence motion on a video.
• Outside Iowa: treat Amyda as persuasive authority at most and check your own jurisdiction's authentication and duplicate rules.
• Defense counsel: tie any deepfake objection to specific evidence about the video, not to general concern about AI. This is our suggestion, not a standard stated by the court.
• Prosecutors: lay the foundation with identifying testimony and circumstances, and document how the file reached law enforcement. This is also our suggestion.

What we did not verify

What we opened: the full Iowa Supreme Court opinion in No. 24-1563 at the Primary link. Every quotation on this page comes from it, with page cites to the slip-opinion page headers.

What we did not open: the proposed Rule 901(c) agenda book, In re M.S., Manning, the other cases the court cites, the current text of Iowa Rules 5.901 and 5.1003, or any docket entry after filing, including a rehearing petition.

What we refuse to claim: that the decision sets a national standard, a forensic test for deepfakes, or a rule that expert evidence is or is not required; anything about the current status of proposed Rule 901(c); or anything about In re M.S. beyond Iowa's description of it.

Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.

Related briefings

AI Regulation News hub | Pillars: North America AI regulation, AI court rules, Attorneys

• Arizona Court Vacates Sentence Over AI Victim Video
• A Proposed Federal Rule Would Put AI-Generated Evidence Through the Same Reliability Gate as Expert Testimony
• Iowa now gates the AI tools its judges may use, and forbids deferring to what they produce
• Virginia Law Now Requires a Human Decision-Maker in Criminal Justice AI Decisions