Public Safety & Emergency ServicesIssue 635 records
Public Safety
This public-safety edition finds accountability being written down after the fact. Anchorage passed a modest ordinance codifying police surveillance and AI rules for tools already in use, and a federal class action alleges Axon’s license plate data practices lack required California disclosures. England and Wales’ Civil Justice Council recommends AI declarations for witness and expert evidence, while a federal judge declined sanctions for AI-drafted citation errors yet disregarded the corrected authorities. In corrections, Michigan’s classification pilot starts Dec. 1 with officials stressing it is not AI and staff decide, amid staffing strain. Evidence is mostly reporting and recommendations, not measured outcomes.
- Evidence records
- 5
- Cross-source patterns
- 2
- Evidence classes
- 4 independent reporting1 standards or public-body guidance
- Outcomes
- 2 emerging2 mixed1 cautionary
- Source freshness
- 5 new this fortnight
- Research completed
- 2026-11-09
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Responsibility for AI-assisted work is being kept with named people, with records to prove it
The CJC would keep responsibility with legal representatives and add declarations where AI generates or influences witness and expert evidence. The N.D. Cal. order says verification duties remain with counsel regardless of AI use or delegation, and penalised the error even without sanctions. Michigan DOC says its classification score is reviewed by staff who make the final decision. In courts and corrections alike, the control offered is human accountability, but only the court cases describe records or consequences that would let it be tested.
Operating questionFor each AI-assisted court or corrections output we use, who is the named accountable person, and what record shows what they checked?
Supporting evidenceCivil Justice Council (England and Wales), as summarised by Lewis Silkin LLPU.S. District Court, N.D. California (order summarised by EDRM / Michael D. Berman)Michigan Department of Corrections with Recidiviz (reported by WCMU Public Media)
Oversight of policing technology is arriving through statute, litigation and council votes after systems are already running
Anchorage members said ALPR, real-time crime center video and AI case-work programs were in use before the Assembly legislated, and the passed ordinance largely codifies existing APD policy while adding annual reporting and approval of materially expanding contracts. A new federal class action alleges Axon’s ALPR data use lacks disclosures California law requires, an allegation not yet adjudicated. Both show governance being applied to established operations, with the strength of the result still uncertain.
Operating questionWhich of our deployed surveillance tools lack a published policy, reporting requirement or contract trigger for expansion?
Supporting evidenceAnchorage Assembly and Anchorage Police Department (reported by Anchorage Daily News)Hausfeld and Gibbs Mura (plaintiffs’ counsel); Axon Enterprise (defendant); reported by Davis Vanguard
Evidence ledger
Full record · every source keeps its link and limitations
Federal class action alleges Axon collects and uses California drivers’ license plate data without the disclosures the state ALPR privacy law requires
Davis Vanguard reports that a proposed federal class action, Ryan Mun et al. v. Axon Enterprise, filed by Hausfeld and Gibbs Mura, alleges Axon collected, used and shared personal information of millions of California drivers through its ALPR network without the disclosures that California’s ALPR Privacy Act (Civil Code 1798.90.5 to 1798.90.55) requires. Citing Reuters, it says the suit seeks more than $5 million and an injunction. Plaintiffs contend Axon is both an ALPR operator and an end user under the statute and must publish a usage and privacy policy covering collection, use, retention and sharing; they allege its use of customers’ ALPR data is not adequately disclosed. The article stresses the allegations are unadjudicated and notes a separate CAIR/EFF suit against San Jose, where the judge signalled preliminary agreement with parts of the city’s position.
Why it matters, evidence and limitations
- Why it matters
- Agencies buying ALPR as a service inherit questions about what the vendor itself does with their read data. The case tests whether a vendor, not only the police customer, carries statutory disclosure duties.
- Evidence and measured results
- Single secondary news report of a complaint; no court ruling, no vendor response captured, and no measured effects. Allegations only.
- Limitations and uncertainty
- Unadjudicated allegations; filing date and complaint details not independently verified; California-specific statute; outcome may not transfer to other states.
England and Wales Civil Justice Council final report recommends AI declarations for witness statements and expert reports and a Practice Direction, but no new rules for lawyers’ drafting
Lewis Silkin summarises the CJC’s final report on AI in preparing court documents. The CJC recommends keeping responsibility with legal representatives and their regulators, supplemented by a regularly updated Practice Direction covering legal representatives and litigants in person; urgent consideration of guidance for litigants in person; a CPR amendment requiring a witness of fact to declare that AI used in preparing a statement has not generated or influenced its substantive content; and a declaration by experts where AI generates or influences their report, with methodology disclosed. It proposes no new declaration rule for pleadings, skeleton arguments or advocacy, and a functional rather than technical definition of AI use. The firm warns that “influence” may be hard to bound and could deter efficient uses. The recommendations are not yet in force.
Why it matters, evidence and limitations
- Why it matters
- Court systems weighing AI rules can compare a risk-based model that targets evidence integrity (witness and expert material) rather than all drafting. U.S. courts differ, so transferability is limited to design lessons.
- Evidence and measured results
- Secondary summary of a final report; recommendations only; no data on prevalence of AI misuse or effects of any rule.
- Limitations and uncertainty
- Recommendations not adopted; civil jurisdiction; commentary by a law firm; U.S. state courts follow different rule paths.
Michigan prison classification pilot starts Dec. 1 amid staffing crisis; MDOC says the score does not use AI and staff make the final call
WCMU reports that from Dec. 1 the Michigan Department of Corrections will pilot a new classification system at select prisons, created with the nonprofit Recidiviz, which was hired to examine historical prison data, hold focus groups and talk with employees. MDOC cites over 300 officer assaults by inmates and over 500 assaults on other prisoners in 2025. After initial coverage assumed AI was involved, MDOC and Recidiviz said AI is not used in classification; MDOC’s Nov. 5 statement says the tool calculates an assessment score that staff review before a final decision. A St. Louis Correctional Facility employee says misclassification has contributed to assaults there, and the report describes staffing vacancy rates over 30 percent at multiple facilities. Pilot sites are not named.
Why it matters, evidence and limitations
- Why it matters
- Shows how a corrections analytics change can be mislabelled as AI, and how staff voices and staffing conditions shape whether a classification tool is trusted. Useful for corrections buyers separating AI features from rules-based scoring.
- Evidence and measured results
- Public-radio reporting with agency statements; no outcome data, validation method or pilot sites published in the article.
- Limitations and uncertainty
- Pilot not started; outcome claims are agency expectations; Recidiviz describes AI use elsewhere in its products for meeting transcription; Michigan-specific.
Federal court declines sanctions for AI-drafted motion with wrong citations but refuses to consider the replacement authorities and warns counsel
EDRM summarises Farhat v. Cutera, Inc. (N.D. Cal. Oct. 28, 2026). Defendant’s motion to dismiss cited cases the court could not locate; the court issued an order to show cause. Counsel’s declaration said a generative AI drafting tool (Legora) produced the first draft, she checked authorities in Westlaw and corrected errors, but an earlier uncorrected version was converted to PDF and filed. She supplied a redline. The court discharged the order and declined sanctions at this time, but disregarded the inaccurate citations and refused to consider the replacement authorities, and warned that counsel must review the final document and verify authorities exist regardless of AI use or delegation, with future errors risking Rule 11(c) sanctions.
Why it matters, evidence and limitations
- Why it matters
- Illustrates a calibrated judicial response: candor and correction avoided sanctions, yet there was still a practical penalty. Relevant to court rules and to prosecutors or public defenders using drafting tools, with limited transfer because this was a civil case.
- Evidence and measured results
- One case summarised by a practitioner blog; no data on frequency of such errors.
- Limitations and uncertainty
- Single order; N.D. Cal. practice; civil commercial litigation; not criminal courts or prosecutors; outcome could differ before another judge.
Anchorage Assembly passes 11-1 a modest ordinance putting police surveillance and AI technology rules into city code, with annual reporting and Assembly approval for expansions
ADN reports that the Anchorage Assembly approved 11-1 an ordinance adding rules on police surveillance and new law enforcement technologies to municipal code. A stricter August proposal lost to a compromise from the mayor’s administration that largely codifies existing APD internal policy and leaves the department more latitude as tools change. It requires an annual administration report to the Assembly covering uses such as drones and license plate readers, sets sanctions in code for rule violations, and requires Assembly approval of any vendor contract that would materially expand surveillance capabilities regardless of amount. A related ad hoc committee will evaluate APD policies. Members said ALPR, real-time crime center video and AI case-work programs were already in use before the Assembly legislated, and described a contentious process.
Why it matters, evidence and limitations
- Why it matters
- A concrete local template: an expansion-approval trigger independent of dollar value, annual use reporting and sanctions in code. It also shows oversight arriving after deployment and in a diluted form.
- Evidence and measured results
- Newspaper account of a council vote; no evaluation of effects; ordinance text not reviewed.
- Limitations and uncertainty
- Alaska municipal context; compromise version is weaker than the original proposal; effects unknown; political reporting.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Independent reporting
- Independent reporting with attributable sources but without a formal evaluation design.
- Standards or public-body guidance
- Normative or advisory guidance from a standards body or public institution.