The Shield Summary Newsletter — June, 2026
This month's issue also addresses the legal landscape facing local law enforcement on the question of ICE detainers — a topic that has moved rapidly from policy debate to active litigation. Four simultaneous legal fronts are now live: a federal lawsuit against the City of Boston challenging its detainer policy, the Sanctuary City Elimination Act moving through Congress, ongoing circuit court development of Fourth Amendment detainer liability, and daily operational decisions being made at agencies without adequate written policy guidance. Sam Hall, Esq. of Crivello, Nichols & Hall, S.C. addresses the complete legal framework on June 25. Registration is $49 at shieldpst.org.
A newly published ACLU study confirms what legal scholars have warned for months: AI-generated police reports were rated by experienced supervisors as "substantively and significantly worse on accuracy" than officer-written reports. The implications for officers who sign off on AI-generated reports — without understanding what the tool inserted, changed, or fabricated — are serious and extend to criminal exposure, civil liability, and evidentiary challenges at trial. Section 06 provides an overview of the eight specific risks every officer and agency needs to understand before using any AI report-writing tool.
Finally, our Use of Force in Transition Symposium — July 13–15 in Green Bay, Wisconsin, co-hosted with Crivello, Nichols & Hall, S.C. — is six weeks away. Space is limited and seats are filling. Details and registration are at shieldpst.org · $695 per attendee.
Decision Watch & the Use of Force Liability Frontier
The close of the Supreme Court's term brings the most anticipated Fourth Amendment decision since Carpenter v. United States in 2018, while new federal court rulings continue to develop the post-Barnes v. Felix use-of-force landscape.
Okello Chatrie was convicted of a 2019 bank robbery near Richmond, Virginia, after law enforcement served a geofence warrant on Google directing it to search its location database — spanning hundreds of millions of accounts — and disclose the identity of every device within a defined radius of the bank during a one-hour window around the robbery. The en banc Fourth Circuit affirmed the conviction in a single unsigned sentence, accompanied by nine concurring and dissenting opinions totaling over 100 pages, with judges offering a range of approaches to the constitutional question but producing no majority rationale. The Fifth Circuit reached the opposite conclusion in a parallel case, holding geofence warrants categorically unconstitutional. The Supreme Court granted certiorari in January 2026 to resolve the split. Oral argument on April 27, 2026, revealed a divided Court; legal analysts assessed the justices as most likely to issue a narrow ruling clarifying scope and particularity requirements rather than a categorical holding in either direction.
Three outcomes are possible. Agencies should prepare for each now, before the decision is issued.
Practical Impact: Regardless of outcome, Chatrie will impose new warrant-application standards that agencies and prosecutors must incorporate before the next geofence application is prepared. Shield PST will publish a same-day analysis at shieldpst.org when the decision issues. Note: Google moved its Location History data to on-device storage in July 2025, which will materially affect the availability of Sensorvault data for future warrants even under a favorable ruling.
Officer Kenton Talton of the Winter Park, Florida Police Department fatally shot Daniel Knight, 39, less than two minutes after arriving at Knight's niece's wedding reception in response to a 911 call describing a man who was "irate and violent." During the confrontation, Knight punched Officer Craig Campbell and then became involved in a physical altercation with Talton before Talton fired seven shots. Mellisa Cruz, Knight's fiancée and the mother of his two younger children, sued Talton personally for wrongful death, as well as the city's police chief, alleging the city's negligence in hiring and supervising Talton.
The federal district judge ruled that the wrongful death suit against Talton personally may proceed — finding that sovereign immunity did not protect the officer given the allegations — and that a Fourth Amendment excessive force claim may also go forward. The judge found that the use of deadly force would not have been objectively reasonable under the circumstances as alleged: the precipitating offense was relatively minor, and the level of force deployed was disproportionate to the threat that actually existed at the moment of the shooting. A second officer, Campbell, was dismissed on sovereign immunity grounds. The ruling was based on the allegations alone; evidence has not yet been reviewed in a full trial proceeding.
Practical Impact: Courts applying the Barnes v. Felix totality-of-circumstances framework will examine the full sequence of events — including what brought officers to the scene, what the precipitating offense was, and whether the level of force deployed was proportionate to the threat that actually materialized. A 911 call describing someone as "irate and violent" does not by itself establish that deadly force was objectively reasonable when the encounter unfolds differently. Officers must document, contemporaneously, what they personally perceived at each decision point — not what the dispatch description said before they arrived.
ICE Detainers & Local Law Enforcement: The Constitutional Landscape
The question of whether a local law enforcement agency may — or must — honor a federal immigration detainer has never been more legally contested. Circuit courts have established a clear Fourth Amendment framework; Congress is pushing in the opposite direction; and the federal government is actively suing cities that refuse. Every agency that operates a jail needs a written, counsel-reviewed detainer policy before the next request arrives.
Form I-247A is an administrative document — not a judicial warrant. ICE issues it to request that a local jail hold a person for up to 48 hours past their lawful release date so federal agents can arrange transfer to federal custody. ICE itself acknowledges in its own guidance that detainers are requests, not orders, and impose no mandatory legal obligation on local agencies. The constitutional problem is direct: extending a person's custody beyond their lawful release date is a new seizure under the Fourth Amendment — one that requires independent legal authority. A civil administrative detainer provides none.
The First Circuit held that detaining an individual solely on a civil immigration detainer — without independent probable cause or a judicially-issued warrant — constitutes an unlawful seizure under the Fourth Amendment. The fact that ICE requested the hold does not provide the independent legal authority the Constitution requires for a new detention. Local agencies that extend custody solely on the I-247A, without a judicial warrant, face § 1983 liability for the unlawful detention.
Practical Impact: In First Circuit jurisdictions, honoring a civil detainer without a judicial warrant is a documented constitutional violation. Good-faith reliance on the administrative I-247A form as independent legal authority has been rejected in this circuit. A judicially-issued warrant, supported by probable cause, is required before any custodial extension.
The Ninth Circuit held that extended detention of persons based on suspected immigration status, without independent legal authority, constitutes an unlawful Fourth Amendment seizure. The court's reasoning — that immigration status alone provides no authority for a local officer to extend custody — applies directly to civil detainer compliance. Agencies in the Ninth Circuit that hold individuals on I-247A requests without judicial authorization face the same constitutional exposure as those who held individuals based solely on apparent immigration status.
Practical Impact: Combined with Morales, this decision means that agencies in at least two major circuits — covering approximately half the country — face documented Fourth Amendment exposure for civil detainer holds conducted without judicial warrants. The majority of civil detainer lawsuits resulting in significant settlements — including the Los Angeles County consent decree requiring judicial warrants before any detainer hold — trace their constitutional theory directly to this body of law.
The Tenth Amendment anticommandeering doctrine — established in Printz v. United States, 521 U.S. 898 (1997) — prohibits the federal government from compelling state and local law enforcement officers to administer federal regulatory programs. Federal courts have consistently held that the federal government cannot mandate local participation in immigration enforcement. Agencies that adopt written policies requiring officers to honor all civil detainers, as a matter of course and without independent legal authority, may be operationalizing a federal mandate that the Constitution does not permit. This is not a theoretical defense — it is an active argument in federal courts, including the Boston litigation described below.
The Trump Department of Justice filed a federal lawsuit against the City of Boston challenging its Trust Act, which limits police cooperation with ICE to criminal matters and bars officers from honoring civil immigration detainers. A federal judge heard oral arguments on May 13, 2026. The government argued that local agencies are legally required to honor detainers — a position that prompted visible skepticism from the bench. Judge Leo T. Sorokin pointed to the 2017 Massachusetts Supreme Judicial Court decision prohibiting local officers from holding persons on civil detainers under state law, and asked the federal prosecutor directly: "You want me to order them to violate state law?"
On May 28, 2026, Judge Sorokin dismissed the DOJ's case entirely. His ruling was sweeping: "In Massachusetts, there is simply no source of authority empowering Boston police officers to do what the United States would like them to do." He held that even without the Trust Act, Massachusetts state law — as interpreted by the Supreme Judicial Court in 2017 — already prohibits Boston officers from holding persons on civil immigration detainers. Because striking down the Trust Act would not give the federal government the relief it sought, Sorokin found the DOJ lacked standing to bring the case. The DOJ lost similar challenges to sanctuary laws in Colorado, Illinois, and New York during the same period.
The dismissal is a significant win for local agencies operating in states with existing sanctuary law authority. But its force is limited to those jurisdictions: agencies in states without analogous state law protections face a different legal environment. The practical principle remains constant in all jurisdictions — the I-247A alone is not sufficient legal authority for a custodial extension, regardless of the federal political environment.
Introduced in Congress in April 2026, the Sanctuary City Elimination Act would: (1) treat local officers who honor detainers as federal agents for purposes of protection from state prosecution; (2) authorize states to sue sanctuary jurisdictions for damages caused by persons released under sanctuary policies; and (3) tie federal funding to detainer compliance. The bill has not been enacted. Its introduction signals the direction of federal legislative pressure and the potential for statutory resolution — in either direction — of questions courts have been developing for a decade.
If your agency honors civil detainers without a judicial warrant: Exposure to § 1983 liability for unlawful detention of any person held solely on Form I-247A. Circuit authority in the First, Ninth, and other circuits supports this claim. Good-faith reliance on the I-247A as independent legal authority has been rejected in multiple circuits. The safest standard in every jurisdiction: require a judicially-issued warrant before extending any custody.
If your agency refuses all detainer cooperation: Potential exposure to loss of federal Byrne JAG and other formula grant funding under conditions attached to federal appropriations. Risk of DOJ enforcement action. Political and operational consequences from federal law enforcement partners and elected officials.
The Practical Resolution — Five Points Every Agency Needs: (1) A written detainer policy, reviewed and approved by agency counsel. (2) A determination of what your state law independently authorizes — this varies jurisdiction by jurisdiction and must be confirmed, not assumed. (3) A requirement of a judicially-issued warrant before any custodial extension — the administrative I-247A is not sufficient. (4) Staff training on the policy, documented in training records. (5) Supervisory authorization required for each detainer decision, documented in the record. Sam Hall, Esq. provides the complete legal and operational framework on June 25.
Register at shieldpst.org · $49 per attendee.
What Agencies Are Paying — and Why
Two significant developments from May 2026 illustrate recurring patterns in law enforcement civil liability: the cover-up that transforms a defensible incident into an indefensible one, and the force decision that comes under scrutiny when the facts on the ground differ from the dispatch description that prompted the response.
Ronald Greene, 49, died in the early morning hours of May 10, 2019, following a vehicle pursuit and violent roadside arrest by Louisiana State Police officers near Monroe. Officers punched, kicked, and applied stun guns to Greene while he was on the ground; troopers then left him lying face down, unattended, for more than nine minutes without rendering aid. LSP initially told Greene's family he had died in a crash. In 2021, the Associated Press obtained body-worn camera footage that directly contradicted that account and documented a sustained violent assault following the stop. Multiple officers subsequently faced criminal charges, though most prosecutions failed; two officers ultimately pleaded no contest to misdemeanor battery. A tentative settlement was reached May 13, 2026 — $4.8 million from Louisiana State Police and $50,000 from the Union Parish Sheriff's Office, for a total of $4.85 million — pending approval by the Louisiana Legislature. The settlement resolves a federal wrongful death lawsuit filed in 2020 by Greene's daughter.
Agency Lesson: The falsification of incident reports, delayed disclosure of body-worn camera footage, and the initial misrepresentation of the cause of death transformed what was a use-of-force incident into a criminal and civil catastrophe of an entirely different magnitude. The cover-up cost exponentially more than the incident ever would have. Every officer who certifies a report, and every supervisor who reviews one, should understand that body-worn camera footage is preserved, produced in discovery, and will be compared against every written representation the agency makes. Post-force documentation is not an administrative obligation — it is a constitutional and, in many jurisdictions, a criminal one.
Officer Kenton Talton shot and killed Daniel Knight, 39, less than two minutes after arriving at a wedding reception in response to a 911 call reporting an "irate and violent" man. During the encounter, Knight struck one officer and was in a physical altercation with Talton before Talton fired seven shots. Mellisa Cruz, Knight's fiancée, filed a wrongful death suit against Talton personally and a negligence claim against the city. A federal judge ruled the personal wrongful death suit against Talton may proceed — sovereign immunity did not provide protection given the allegations — and that a Fourth Amendment excessive force claim is also viable. A second officer, Craig Campbell, was dismissed on sovereign immunity grounds. The case proceeds to discovery and trial.
Agency Lesson: A 911 caller's description of a subject as "irate and violent" does not predetermine the constitutional analysis of what force was justified when officers arrived. Under Barnes v. Felix, courts evaluate the full sequence — what officers observed upon arrival, the severity of the offense at issue, the subject's actual behavior at the moment force was used, and whether lesser means were available and considered. Officers must be trained to make real-time threat assessments independent of dispatch characterization, and to document those assessments with specificity in their use-of-force reports.
Suicide Prevention, Federal Investigation & the AI Compliance Deadline
Three developments — a state correctional system's proactive suicide prevention reforms, a new DOJ CRIPA investigation into a women's prison, and a technology compliance deadline arriving June 30 — frame the June correctional law update and carry direct implications for jail administrators, risk managers, and agency counsel.
In March 2026, the Massachusetts Department of Correction announced new suicide prevention procedures, clinical coordination protocols, and safety operations across the correctional system following an audit ordered after six prisoner suicides in 2025. The reforms include enhanced mental health screening at intake, mandatory supervisory review of all cells housing individuals on suicide watch, and a coordinated clinical response protocol tying custodial decisions to documented clinical assessment findings. For correctional administrators, the significance of this announcement lies not only in its content but in its timing: the reforms were adopted proactively, before litigation compelled them.
Under Estelle v. Gamble, 429 U.S. 97 (1976), and its progeny, deliberate indifference to serious medical and psychiatric needs constitutes an Eighth Amendment violation. Courts apply the same standard to suicide prevention: when an agency is aware of the risk and fails to implement adequate screening and precaution protocols, the gap between known risk and agency action is the core of a deliberate indifference claim. Proactive reform — documented before an incident and demonstrably implemented — is the most defensible posture an agency can occupy. Agencies waiting for litigation to compel reform are building the plaintiff's damages case one day at a time while they wait.
On March 26, 2026, the Department of Justice announced a civil rights investigation into whether Maine engages in a pattern or practice of violating the constitutional rights of female prisoners incarcerated at Maine Correctional Center in Windham, pursuant to the Civil Rights of Institutionalized Persons Act, 42 U.S.C. § 1997. The investigation addresses conditions of confinement, medical and mental health care, and use of force. Maine joins Colorado, whose Department of Corrections and Division of Youth Services came under CRIPA investigation in December 2025. The geographic distribution of open CRIPA investigations now spans New England, the Mountain West, and the South. Correctional administrators should not assume that geographic distance from prior investigations confers protection from scrutiny — the DOJ's Civil Rights Division investigates based on indicators, not proximity.
ADA Compliance Obligations Do Not Change With Enforcement Climate: Title II of the ADA applies to county jails and state correctional facilities regardless of who controls the Department of Justice or whether a federal investigation is active. Shield PST's ADA in the Jails symposium — coming this fall — provides correctional commanders, jail administrators, and agency legal counsel the legal framework, policy tools, and practical guidance needed to meet ADA obligations independent of the federal enforcement environment.
Register here · $795 per attendee · Location to be announced · shieldpst.org
Colorado's Artificial Intelligence Act takes effect June 30, 2026. The Act applies to state and local government entities that deploy AI systems in "high-risk" decisions, which specifically includes law enforcement identification tools, public safety decisions, and criminal justice risk assessments. Compliance obligations include: (1) documentation of the AI system's training data and known limitations; (2) a pre-deployment impact assessment; (3) public notice of AI tool use; and (4) individual recourse mechanisms for persons affected by AI-assisted decisions. Law enforcement agencies in Colorado deploying facial recognition, predictive policing tools, or AI-assisted report writing must confirm compliance before June 30. A similar statute — the Texas Responsible AI Governance Act — took effect January 1, 2026. More state legislation is moving through legislatures across the country; agencies in all states should be tracking this regulatory development actively.
When the AI Writes the Report — and the Officer Signs It
AI report-writing tools have proliferated rapidly across American law enforcement. A study released by the ACLU in May 2026 found that AI-generated reports were rated by experienced supervisors as "substantively and significantly worse on accuracy" than officer-written reports — moving a report from roughly the 50th to the 36th percentile on perceived accuracy. The officer who signs that report owns every word of it.
AI report-writing tools generate text from body-worn camera audio transcripts or officer voice input. The officer then reviews and certifies the report as accurate. In every jurisdiction, that certification carries legal weight. The AI's authorship is irrelevant to the certifying officer's responsibility. If the report contains a factual error — a wrong time, a misquoted statement, a description of conduct that did not occur — and the officer signed it, the officer certified a false record. "The AI wrote it" is not a defense to a false report allegation, a disciplinary charge, a Franks hearing, or a criminal prosecution for submitting a false official document. The report is the officer's.
AI language models are documented to "hallucinate" — generating plausible-sounding but factually incorrect content. In a police report context, hallucination can produce: witness statements the witness never made; physical descriptions of subjects that do not match BWC footage; times, distances, or sequences inconsistent with the actual encounter; and articulations of probable cause elements that the officer never personally observed. When the report is compared against body-worn camera footage in discovery, inconsistencies generated by AI hallucination look identical to intentional falsification. Defense counsel will not make a distinction between the two. The officer must be able to account for every factual assertion in the report because opposing counsel will hold them to every word of it under oath.
When an AI tool generates a report from a BWC transcript and the officer reads and edits it, the officer's independent recollection of the event is influenced by what they read. If the AI version contains details that differ from what the officer actually experienced, reading that version begins to overwrite the officer's authentic memory. By the time the officer testifies at trial — months or years later — it may be impossible to cleanly separate what was personally observed from what the AI described. The ACLU has identified this as a civil liberties problem that is "not even theoretically solvable" — and defense attorneys are already trained to exploit this ambiguity in cross-examination. An officer who cannot testify with specificity to what they personally perceived, independent of the report, is a compromised witness.
Police reports are foundational documents in criminal prosecutions. Defense counsel can — and increasingly will — challenge AI-generated reports on authentication grounds: Who authored this document? Can the officer testify to every factual assertion as based on personal knowledge? Was the BWC audio accurately transcribed before the AI processed it? What version of the AI model was used, and is that model's output reproducible? Proprietary AI systems — particularly vendor tools with protected algorithms — may prevent defense counsel from examining how the report was generated, raising potential due process concerns that courts are beginning to examine. Admissibility challenges to AI-generated reports have not yet produced a definitive circuit-level ruling, but the groundwork is being laid in discovery proceedings in jurisdictions nationwide.
If it is discovered — through discovery, a public records request, or vendor disclosure — that an officer routinely uses AI report-writing tools, that information may constitute Brady or Giglio material requiring disclosure in every prosecution in which the officer is a witness. Prosecutors who are unaware that their officers use AI tools may unknowingly fail to make required disclosures, creating Brady violations with consequences for pending and completed cases alike. At least one public defender's office has already begun filing discovery demands specifically requesting disclosure of AI tool use in report drafting for all officer witnesses. Every prosecutor who relies on officers who use AI tools needs to know — before trial — what tool was used, on which reports, and whether any discrepancies between AI-generated text and BWC footage exist.
When AI-generated language is incorporated into a warrant affidavit — or when an AI-generated incident report is referenced in one — the officer is swearing to the truth of AI-generated content. A Franks v. Delaware, 438 U.S. 154 (1978) hearing allows a defendant to challenge the truthfulness of statements in a warrant affidavit. If an AI tool generated language asserting probable cause elements that the officer did not personally observe, or that are inconsistent with BWC footage, a successful Franks challenge results in suppression of all evidence obtained pursuant to the warrant — regardless of what was actually found. This is the direct and logical consequence of using AI-generated text in sworn legal documents without rigorous, line-by-line, personal-knowledge verification.
Most agencies do not yet have written AI report-writing policies. Officers are using consumer tools — ChatGPT, Claude, Gemini, and others — without authorization, without vendor data security agreements, and without any understanding of where their agency's sensitive law enforcement information is being transmitted and stored. In December 2025, the San Diego Police Department issued a general order prohibiting all unauthorized AI tools in report writing, following the passage of California Senate Bill 524, which requires agencies to formally adopt AI policies, disclose tool use, and document who used the tool, who edited the report, and what footage was used. Officers using unauthorized tools at agencies without a written policy face disciplinary action. Agencies without a policy face both the liability exposure of unauthorized use and growing statutory exposure under emerging state AI governance legislation.
Utah required disclosure of AI use in police reports before October 2025. California's Senate Bill 524 now mandates formal agency AI policy adoption, disclosure of tool use, and documentation of who used the tool, who edited the report, and what footage was the source. Colorado's Artificial Intelligence Act takes effect June 30, 2026 — imposing pre-deployment impact assessments, public notice requirements, and individual recourse mechanisms on government AI use in high-impact decisions, expressly including law enforcement. The Texas Responsible AI Governance Act took effect January 1, 2026. Additional state legislation is active in multiple other states. Agencies that deploy AI report-writing tools without written policy, officer training, supervisory review, and state-law compliance are building regulatory and civil liability exposure simultaneously.
The Bottom Line: AI report-writing tools may offer convenience. They do not offer legal protection. An officer who certifies an AI-generated report has certified every word of it — including words the AI inserted, confabulated, or fabricated. The officer's memory, credibility, and professional standing are on the line in every proceeding that follows. The agency that deployed the tool without a policy, without training, and without supervisory review has created the conditions for Monell liability if an AI-generated report is ever shown to be the proximate cause of an unlawful arrest, a failed prosecution, or a constitutional violation.
Shield PST's AI-Assisted Police Report Writing virtual workshop provides the legal framework, practical guidance, and documented training that officers and supervisors need to use these tools safely — or to make an informed decision not to. Details at shieldpst.org.
9:00 AM PDT / 12:00 PM EDT
Nationwide
shieldpst.org
What You Will Learn:
- • The Form I-247A framework — what it is, what it requests, and what it does not legally authorize
- • Fourth Amendment seizure doctrine applied to civil detainer holds — circuit authority and the national exposure map
- • The anticommandeering doctrine and what it means for local agencies asked to enforce federal immigration law
- • State law authority — a jurisdiction-by-jurisdiction overview of what your state law does and does not authorize
- • The Boston Trust Act litigation — current posture, likely outcomes, and national implications
- • Liability exposure for agencies that honor detainers without a judicial warrant
- • Liability exposure for agencies that limit or refuse detainer cooperation
- • The five-point policy and training framework every agency needs before the next detainer request arrives
- • Documentation and supervisory authorization requirements for every detainer decision
2026 Shield PST Training Calendar
| Date | Event | Location | Type |
|---|---|---|---|
| June25, 2026 |
ICE Detainers & Local Law Enforcement
Sam Hall, Esq. · 9:00 AM PDT / 12:00 PM EDT · $49 · Register here
|
Online | Webinar |
| July13–15, 2026 |
Use of Force in Transition Symposium
Co-hosted with Crivello, Nichols & Hall, S.C. · $695 · shieldpst.org
|
Green Bay, WI | Symposium |
| September2026 |
ADA in the Jails: The Fragile Inmate
$795 · Register here · shieldpst.org
|
TBD | Symposium |
| Nov/Dec2026 |
Advanced Internal Affairs
$795 · shieldpst.org
|
Henderson, NV | Symposium |
Webinars: $49 per attendee · Symposiums from $695 · Register at shieldpst.org
