AI-Generated Song Played in Murder Trial Raises Questions About What Digital Evidence Can Prove

An AI-generated song played during Caleb Flynn's murder trial has highlighted how quickly artificial intelligence is becoming part of the evidence examined in criminal cases.

Flynn, a former American Idol contestant and Ohio pastor, was found guilty on 29 September of murdering his wife, Ashley Flynn, who was killed at the couple's home in Tipp City, Ohio, on 16 February 2026. The jury also convicted him on eight other charges, and he is due to be sentenced on 5 October.

The song was one element of a broader prosecution case that included material from Flynn's devices, more than 100,000 messages exchanged with Alleigha Botner, and testimony about his relationship with her. Ohio Bureau of Criminal Investigation special agent Joseph Wilhelm also testified about digital material recovered during the investigation.

The AI Song Was Part of a Larger Digital Trail

The prosecution's case did not depend on the song alone. Investigators examined digital material from Flynn's devices, while jurors heard evidence about more than 100,000 messages exchanged between Flynn and Botner. Botner also testified about the relationship and explained the lyrics of songs associated with Flynn.

Wilhelm testified that an AI music-generation application had been uninstalled from Flynn's phone before Ashley's death, while investigators found songs on the device. Reports from the trial said the lyrics in the songs could be matched with material in Flynn's notes.

That context matters. The AI-generated music was not significant simply because artificial intelligence had produced it. Its relevance came from its connection to a particular device, a particular relationship, and other material presented to the jury.

Ohio Law Already Covers Digital Authentication

The trial did not establish a new category of evidence specifically for AI-generated content. Ohio Evidence Rule 901 says material must be authenticated with evidence sufficient to support a finding that it is what its proponent claims. The rule allows authentication through testimony from someone with knowledge, as well as through distinctive characteristics and surrounding circumstances.

Ohio courts have applied that framework to digital material before the rise of generative AI. In State v Perenkovich, for example, the Ohio Fifth District Court of Appeals said photographs of text messages could be authenticated through testimony from a witness with personal knowledge. The court distinguished questions about whether evidence is genuine from arguments about how persuasive that evidence is.

For AI-generated evidence, that distinction could become increasingly important. A court may need to establish that a particular file is authentic without necessarily treating every word, image, lyric, or claim produced by software as factual.

The Human Connection Can Matter More Than the AI

The Flynn case illustrates why investigators may look beyond the content of an AI-generated file. A song generated by software can contain synthetic vocals or music, but investigators can still examine the device on which it was stored, associated files, timestamps, account activity, prompts, notes, messages, and testimony from people who know how it was created.

That creates several separate questions: Is the file genuine? Who had access to it? What human material went into creating it? And what does its existence actually demonstrate? Those questions are different from asking whether AI itself is reliable.

In Flynn's case, prosecutors used the songs alongside other material concerning his relationship with Botner. The jury ultimately found Flynn guilty, while the defence argued that an affair and hostile messages did not themselves establish that he committed murder.

AI Could Make Digital Provenance More Important

The wider issue extends beyond murder trials. As generative AI becomes capable of producing realistic music, images, video, and text, digital evidence may increasingly require investigators and courts to establish not just what a file contains, but how it came into existence.

That does not mean AI-generated material is automatically unreliable. Nor does Flynn's conviction establish a special legal rule for AI evidence. Instead, the case provides a glimpse of how traditional authentication principles could be applied to a new generation of digital material. The technology may change, but the courtroom still has to determine what a piece of evidence is, where it came from, and what conclusions can reasonably be drawn from it.