{"resourceId":"cdt-automated-police-report-drafting-20260330","versions":[{"version":"external-98e92b35ef28123ecfefdb3058d09defcda96a5a59e4b98ad9ba05be7b1af096","resource":{"id":"cdt-automated-police-report-drafting-20260330","title":"CDT: Automated police-report drafting needs disclosure, audit trails, and use limits","organization":"Center for Democracy & Technology","sector":"Civil rights / technology policy (law enforcement)","geography":"United States","publishedAt":"March 30, 2026","publicationDate":"2026-03-30","eventDate":null,"sourceName":"Center for Democracy & Technology","sourceLabel":"Independent civil-rights research brief","sourceUrl":"https://cdt.org/insights/automated-police-report-drafting/","evidenceClass":"independent-research","outcomeClass":"cautionary","topics":["knowledge-work","developers-agents","governance-procurement","data-security","operating-model"],"finding":"CDT’s 30 Mar 2026 analysis of automated police-report drafting (BWC audio/video or field notes → generative narrative) argues reports shape charging, detention, pleas, and sentences, so AI drafts raise civil-rights stakes. It flags transcription errors (citing ~10% word-error rates even in ideal conditions), hallucination/speaker misattribution, automation bias/cognitive offloading marketed as a feature, and weak audit trails—noting some vendors advertise limited logs to “reduce FOIA exposure.” Recommendations: procurement transparency and independent accuracy verification; officer training/certification; restrict high-stakes use until empirical validation; mandatory disclosure plus retention of AI drafts/edit histories (citing CA SB 524 vs narrower UT SB 180); discovery rules covering AI drafts; independent bias audits. Newly relevant after 19 Sep Michigan Draft One / prosecutor-caution coverage: CDT supplies the civil-rights control checklist those pilots still need.","sledRelevance":"Why now: pairs with the stream’s 19 Sep Michigan Draft One and prosecutor-gate pattern—CDT was not previously archived and supplies procurement, discovery, and FOIA/accountability controls U.S. agencies should apply before expanding AI report writing.","evidence":"Policy analysis with cited examples (including bizarre Draft One frog hallucination anecdote) and statutory comparisons (CA SB 524; UT SB 180). Not a new randomized field trial; relies on cited technical literature for WER ~10% and known deployments (Axon, Truleo claims).","architectureImplications":"Interpretation: require edit-history logging, AI-use labels in RMS, and source-to-draft linkage (BWC clip IDs); reject configurations that only store final sign-off.","governanceImplications":"Interpretation: treat FOIA-minimizing audit designs as a procurement red flag; align PD policy with prosecutor discovery expectations before rollout.","securityPrivacyImplications":"BWC audio egress to vendors, prompt retention, and training-data reuse restrictions are central; limited logs impede both oversight and defense rights.","caveats":"March 2026 publication (older than last run) included for new relevance to 19 Sep Michigan wave; advocacy organization perspective; not a multi-site outcome evaluation.","streamIds":["public-safety"],"roles":{"sales":"Sales framing: After Michigan prosecutor caution, CDT reframes the deal as rights-preserving report assist—disclosure, full edit trails, and low-stakes limits—not FOIA-light logging. Sell compliance packaging against CA SB 524-class rules. Stakeholders: chiefs, city attorneys, prosecutors, civilian oversight, unions. Bounded engagement: policy+procurement checklist and discovery tabletop before license expansion. Unsupported: proven time savings or bias-free drafts. Ask whether edit histories survive the full case lifecycle.","engineering":"Engineering framing: Integrate BWC→transcript→draft→officer edit with immutable audit log of every AI token accepted/rejected; label AI assistance in RMS. Prerequisites: CJIS review, prosecutor discovery rules, retention equal to case life. Constraints: vendor products that omit edit history. Security: minimize FOIA-evasive designs; contractually ban training on agency audio. Proof of value: reconstruct officer vs AI authorship for sampled cases; transcription WER tests on local audio.","delivery":"Delivery framing: Implement disclosure labels, draft retention, supervisor sampling, and prosecutor briefing. Dependencies: Axon/Truleo/other vendor features, RMS fields, training curriculum. Ownership: patrol for use; records for retention; prosecutor for discovery posture. Skills: error-hunting over rubber-stamping. Acceptance: AI labeled; drafts retained; no FOIA-minimizing log mode in production. Risks: automation bias; Brady exposure; public trust loss if logs are thin."},"roleDetails":{"sales":{"customerProblem":"Agencies want paperwork relief but face civil-rights, FOIA, and discovery blowback if AI drafts lack provenance.","stakeholders":["Police executive","City/county attorney","Prosecutor","Civilian oversight","Records/FOIA officer"],"discoveryQuestions":["Does the vendor retain full edit history?","Is AI use labeled for discovery?","Are high-stakes offense types excluded?","Who audits differential language impacts?"],"valueHypothesis":"Bounded value is controlled draft-assist with reconstructible authorship—not FOIA exposure reduction.","potentialEngagement":"Procurement red-team and policy gap assessment against CDT checklist and state disclosure laws.","unsupportedClaims":["Guaranteed time savings","That officer sign-off cures hallucinations","Bias-free narrative generation"]},"engineering":{"fit":"BWC/field-note draft assist with mandatory provenance—not silent autofill of official narratives.","architectureIntegration":"Source media ID → transcript → AI draft → tracked edits → labeled RMS narrative → retention store.","prerequisites":["CJIS/privacy review","Edit-history capable product","RMS AI flag","Prosecutor input"],"constraints":["Vendors advertising FOIA-light logs","Noisy BWC audio WER","Mixed state statutes"],"security":"Contract bans on secondary use of audio; encryption; litigation holds on drafts/prompts.","proofOfValue":["Local WER benchmark","Authorship reconstruction drill","Prosecutor acceptance of discovery package"]},"delivery":{"work":"Roll policy, training, labeling, retention job, and monthly supervisor accuracy sample.","dependencies":["Vendor audit features","RMS configuration","Training hours","Union notice"],"ownership":"Patrol commander for field compliance; records manager for retention; legal for discovery.","skillsAdoption":"Train officers to compare drafts to BWC before attesting personal knowledge.","governanceCheckpoints":["Pre-prod audit-log test","30-day sample review","Expand/narrow offense types"],"acceptanceCriteria":["AI disclosure on filings","Draft+edit history retained case-life","FOIA-minimizing mode disabled"],"risks":["Automation bias","Undisclosed AI text in charging packets","Civil-rights disparate framing"]}},"retrievedAt":"2026-09-21T03:10:11Z","enrichedAt":"2026-09-21T03:22:00Z","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Officers need recurring training to catch transcription/speaker errors; supervisors need rubrics that catch incompleteness, not just polish.","procurementImplications":"Write RFPs requiring edit-history export, AI labeling, independent accuracy tests, and prohibition on marketing FOIA-avoidance as a feature.","operatingModelImplications":"Keep human attestation meaningful: policies should forbid filing AI text the officer has not compared to source evidence.","sourceVerification":{"openedUrl":"https://cdt.org/insights/automated-police-report-drafting/","referenceExcerpt":"Remarkably, some vendors even advertise these more limited audit trails as a feature to “reduce FOIA exposure,”","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}