Background
In June 2021, Jersey City police investigated the fatal shooting of Ahmad McPherson on Stegman Parkway. The following day, a confidential informant who had not witnessed the shooting reviewed CCTV footage and identified two men by street names — “Fat Daddy” and “Parkz” — providing Instagram usernames for each. Officers retrieved the Instagram profile photo attributed to Fat Daddy and ran it through the NJ/NY High Intensity Drug Trafficking Areas (HIDTA) Facial Recognition Module. The search returned defendant Tybear Miles as one of ten possible matches — ranked eighth. Police then showed surveillance video and still photographs from the grocery store near the scene to four non-eyewitnesses: a neighborhood resident, defendant’s sister, his ex-girlfriend, and a local man. All four identified defendant from that footage, though none witnessed the actual shooting and no footage captured the moment of the killing.
A grand jury indicted Miles on charges of first-degree murder, second-degree possession of a weapon for an unlawful purpose, and second-degree unlawful possession of a weapon. Defense counsel moved to compel FRT-related discovery, invoking the Appellate Division’s earlier decision in State v. Arteaga, 476 N.J. Super. 36 (App. Div. 2023), which had directed production of thirteen specific categories of FRT material on the facts of that robbery case — including the algorithm’s source code. The trial judge applied Arteaga mechanically and ordered all thirteen items. The State provided only partial discovery: printouts from two FRT searches and a letter from the New Jersey State Police stating it had no record of any search request in the case, leaving the provenance of the search results unexplained. The Appellate Division denied the State’s motion for leave to appeal, and the Supreme Court granted review.
The Court’s Holding
Writing for a unanimous Court, Justice Fasciale held that Arteaga‘s thirteen-item list cannot be applied as a rigid, universal checklist in every case where the State used facial recognition technology. FRT discovery is a case-specific inquiry governed by relevance: whether the requested material tends to prove or disprove a fact of consequence to the defendant’s case. The Court affirmed the trial judge’s order requiring the State to disclose (1) the identity of the FRT tools and materials used in the investigation and (2) information about how the State actually employed those tools — because the first category bears on FRT reliability and the second could support impeachment of the interviewees’ identifications, challenge the quality of the investigation, and raise the possibility of third-party guilt. The Court emphasized that this basic two-category disclosure will, in most cases, constitute the minimum necessary to safeguard a defendant’s right to a fair trial when FRT has been used.
The Court reversed without prejudice the portion of the discovery order compelling production of proprietary FRT materials, including the source code. Relying on the burden-shifting framework from State v. Pickett, 466 N.J. Super. 270 (App. Div. 2021), the Court held that a defendant seeking proprietary algorithmic information must demonstrate a “particularized need” under a four-factor test once the State makes a good-cause showing to shield trade secrets. Here the record was undeveloped — defendant himself acknowledged he did not yet know whether source code or other proprietary data was necessary to his defense — and the trial court had not applied the Pickett framework. The matter was remanded for further proceedings to develop that record before any ruling on proprietary disclosure is made.
Key Takeaways
- Courts must assess FRT discovery requests case by case based on relevance — mechanical adoption of the Arteaga thirteen-item checklist is error.
- In nearly every case where FRT was used, defendants are entitled to at minimum: (1) identification of the specific FRT tool and materials used, and (2) information on how the State deployed those tools against the defendant.
- Proprietary FRT information such as source code is not automatically discoverable; defendants must demonstrate “particularized need” under the Pickett burden-shifting framework, and that analysis requires a developed evidentiary record.
- The State cannot insulate an FRT-driven investigation from discovery simply by declaring it will not introduce FRT evidence at trial — the technology’s role in identifying and targeting a suspect remains relevant to the defense.
Why It Matters
This decision is the first New Jersey Supreme Court ruling on criminal discovery rights when law enforcement uses facial recognition technology, and it arrives as FRT has become a routine investigative tool in departments across the state. By rejecting a one-size-fits-all checklist while simultaneously guaranteeing a baseline of disclosure, the Court charts a middle course between total opacity and reflexive disclosure of proprietary algorithms. Defense attorneys now have clear authority to demand how FRT was used — information critical to challenging tunnel-vision investigations, probing false matches, and identifying alternative suspects — without needing to litigate anew whether FRT discovery is available at all.
The opinion also signals that the Court takes seriously the well-documented reliability concerns surrounding facial recognition — noting expert testimony that FRT systems vary widely in accuracy, perform worse in real-world conditions, and can embed racial and other biases — without yet resolving how those concerns translate into full algorithmic disclosure rights. The remand framework under Pickett means that source-code battles are coming in individual cases, and the outcome of those proceedings will shape whether New Jersey’s broad open-file discovery tradition ultimately extends to the inner workings of commercial FRT systems used by police.