This page is about Asheville: the city’s Real-Time Intelligence Center, approved by Asheville City Council on May 12, 2026, and the statements made for it by Asheville police officials, council members, the mayor, and the vendors, at council and in local press. Each statement is held, quoted from the cited source, beside the dated facts in the record that bear on it. Where a fact is a document, the document is linked; where it is an absence, the date it was looked for is given. The statements and the facts sit side by side so a resident can read both before a meeting.
Crime clearance and the Hillcrest case
The statement. Chief Stepp, May 12, 2026: “This is about improving emergency response, investigative efficiency, transparency and officer accountability.” “Time is often the difference between life and death.” “This is not about software — this is about victims.” (828NewsNow · Annie Dance) The case cited: the March 2025 Hillcrest shooting, in which Flock located the fleeing vehicle “almost instantly” (Capt. Silberman, WLOS).
The record. APD has not published aggregate figures for the program: no count of arrests attributable to the cameras and no response-time data, in any coverage found through Sept 8, 2026. Austin’s program logged 117 million scans against 228 arrests before the city ended it, and Atlanta’s homicide clearance rate fell during its camera buildout (Communities that ended or limited camera contracts).
The audit trail and the July 2025 records refusal
The statement. Stepp, May 12, 2026: “There’s a really robust audit trail built into all of this.” Technology manager James Wingo, Aug 6, 2026: “Our lieutenant does quarterly audits. He just did an audit, and 100% of the time, we have reasons.” (828NewsNow)
The record. APD’s audits are internal and have not been published. A resident requested the city’s Flock audit records in July 2025; the city refused on Sept 9, 2025, citing G.S. 20-183.31(b) and G.S. 132-1.4 (the request and its re-scoped successor). Of the two plate-reader portals APD announced, the Axon portal publishes retention, device counts and hotlist hits; the Flock portal showed an overview paragraph only on Sept 23, 2026. Neither publishes audit results (Promise Tracker, row 10). In Mooresville, a Flock officer’s 31 searches each carried a logged reason and passed review; the reasons were later alleged to be false (Camera network misuse, documented cases). EFF’s review of 11.4 million Flock searches found more than 14 percent justified with the single word “investigation.”
Data ownership and federal sharing
The statement. Stepp, May 12, 2026 (4:47:05): “Axon neither has access to the Fusus or to the system, or retains data. Actually, police department owns the information.” To WLOS: “APD owns the data that we collect, and we have to agree to share that information out, regardless of who requested it.” Deputy Chief Aardema, same coverage: “We control Axon Fusus and we will not be sharing information with federal immigration.” Capt. Moore: “We do not and will not share that information short of a court order.” Axon’s representative, to the Watchdog: “We do not have access. We’ve never had access.” (WLOS · Watchdog)
The record, for Axon. The master agreement the city posted with the contract in August 2026 (Axon’s Master Services and Purchasing Agreement, version 19.0, Exhibit B) says the agency “controls and owns all right, title, and interest in Agency Content.” The same section: “Agency agrees to allow Axon access to Agency Content to (a) perform troubleshooting, maintenance, or diagnostic screenings; and (b) enforce this Agreement.” Usage data about the agency’s system, devices and users is “Non-Content Data,” which the agency agrees Axon may access to “provide, develop, improve, and support current and future Axon products.” The agreement enrolls each agency by default in Axon’s Customer Experience Improvement Program (“By default, Agency will be a participant in ACEIP Tier 1”), under which Axon “may make limited use of Agency Content from all of its customers” to build products, license plate data among its examples. The Fusus appendix: “Axon may collect, use, transfer, disclose and otherwise process Customer Content in the context of … monitoring the Customer’s use of FUSUS systems, and undertaking data analytics.” The city’s opt-out from the program rests on an Axon sales representative’s July 16, 2026 email; Axon’s written confirmation has not been posted (The RTIC Policy). The contract does not state which Fusus tier and what quantities were bought (they are in the unposted quote), a retention rule for Fusus content, or a sharing rule for Axon’s plate readers. Axon states it does not sell personal data, and nothing in the record contradicts that. The terms are quoted in full on the city’s page, annotated; what the same terms did elsewhere is on What Axon Contracts Do.
The record, on sharing. Policy 1703 does not mention immigration; its sharing clause allows disclosure “as permitted by law or with a compelling need to share such information for legitimate law enforcement purposes” (The RTIC Policy). No list of agencies with access to Axon’s plate-reader data has been published. The published sharing limits are the Flock program’s: a 500-mile radius through August 2026, narrowed to four states by the Aug 14 memo (the memo). In the cases documented nationally, immigration searches of a local network were run by another agency in the network rather than requested of the local department (Camera network misuse, documented cases). Aardema, in the same WLOS coverage: “in other areas you see law enforcement agencies using data in ways that would not be appropriate for the Asheville community.”
Search-term blocks
The statement. Wingo, July 8, 2026: “We have blocks on words that have anything to do with immigration, anything to do with reproductive issues.” (WLOS)
The record. In Texas, a deputy’s search of 83,000 cameras for a woman who had an abortion was later reclassified as a missing-person case. The ACLU has documented officers advising each other on what to enter in the reason field (Camera network misuse, documented cases). A block on search terms applies to the words an officer types.
Predictive use
The statement. Deputy Chief Aardema, in the WLOS coverage linked above: the RTIC is “a reactive system, it’s not predictive police.”
The record. In July 2025, Chief Mike Lamb described the department’s approach to the Public Safety Committee using the words “predictive policing,” twice, while presenting crime-reduction figures. To the same committee that month APD said, “APD does not utilize the license plate reader systems at all to alert to immigration violations… They’re turned off.” The alert function remains in the platform as a setting.
Scope: violent crime only
The statement. Aardema: “We have been clear from the get-go that this is a reactive system, it’s not predictive police and the primary focus is going to be on violent crime.” Stepp: “We’re not going to just randomly surveil or monitor people.”
The record. The limits in the approved package are policy, revisable without a council vote (Promise Tracker). Austin’s audit found 10 to 20 percent of searches with no documented purpose; San Francisco police watched protests live through a private camera network; Memphis police surveilled activists in violation of a federal consent decree (Camera network misuse, documented cases).
Facial recognition
The statement. Stepp, on several occasions; Axon’s representative, April 28, 2026: “We also do not use any form of facial recognition.”
The record. The exclusion is in Resolution 26-97, adopted May 12, 2026. A resolution carries no enforcement mechanism and can be changed by a council vote. The policy workgroup’s notes record Stepp’s statement that facial recognition “would require a return to Council for approval” (The RTIC Policy). Axon is testing face recognition on body cameras of the model Asheville bought (What Axon Contracts Do). Cities that have bound the exclusion did so by ordinance requiring a council vote before any new capability activates (Communities that ended or limited camera contracts).
Drones and de-escalation
The statement. Council member Bo Hess, March 25, 2025, on the city’s video at 32:50: “When you actually responsibly deploy technology like drones and real-time intelligence, what you do is you actually de-escalate the risk for officers and our citizens. You’re removing the need to send an officer blindly into a dangerous situation and that lowers the chance of use of force [incidents].” On Durham: “They didn’t throw out public safety, they modernized it.”
The record.
- No evaluation of a drone-as-first-responder program, including Chula Vista’s, has measured use-of-force outcomes; the documented benefit is response speed. The author of the field’s first randomized trial describes the literature as “anecdotal testimonies… hardly any rigorous analyses.” (GovTech · USC)
- The Real-Time Crime Center research institute at Florida State writes that “little empirical work has been conducted on their effectiveness”; the Miami study measured case clearance, not use of force. (FSU · Justice Quarterly)
- The Los Angeles Police Department’s Inspector General found 32.3 percent of Flock stolen-vehicle alerts false in a two-month sample, 161 alerts, each able to trigger a high-risk stop. The Institute for Justice counts 27 documented wrongful plate-reader stops since 2018, with guns drawn in about two-thirds. Aurora, Colorado paid $1.9 million after a misread plate led to four children being held at gunpoint. (NBC LA · IJ · CNN)
- WIRED’s analysis of about 10,000 Chula Vista flights found them concentrated over the city’s lowest-income neighborhoods, with one flight in ten recording no stated purpose; the city lost a four-year records case over the footage. APD’s drones have been used in enforcement of homeless-related offenses (WLOS). (Reason · The Intercept)
- Durham voted 4 to 2 in March 2024 to end ShotSpotter, had no real-time crime center, and in February 2026 its council declined to vote on a proposed one. Durham’s HEART unarmed crisis-response program handled more than 2,000 calls; the city’s violent crime fell 17 percent in 2025. (WUNC · ABC11 · ABC11)
- 2025 recorded 111 line-of-duty police deaths nationally, the fewest since 1943; about a third were traffic-related. (NLEOMF · FBI LEOKA)
- Denver’s STAR unarmed-response program reduced targeted low-level offenses 34 percent in a peer-reviewed evaluation (Science Advances, 2022). Asheville’s Community Responder pilot has 8 grant-funded positions. (Science Advances · Durham HEART report)
Crisis intervention training figures
The statement. Hess, March 25, 2025, about 37:05 on the video: “Over 95% of our APD officers are trained in crisis intervention for mental health, addiction, and de-escalation. And that’s something we should be proud of.”
The record. The last published figure was 90 percent (WLOS, October 2023). No public document giving a figure above 95 percent has been found as of Sept 8, 2026; a records request for the training roster would settle it. The Department of Justice’s CrimeSolutions clearinghouse rates crisis intervention teams “Ineffective” at reducing use of force and arrests; the Memphis-model founders describe CIT as a specialist-team model rather than a training percentage. (WLOS · DOJ CrimeSolutions · Taheri meta-analysis)
The May 12 oversight commitments
The statement. Stepp, May 12, 2026 (5:09:08): “Trust will not be assumed. It will be demonstrated.”
The record, as of Sept 24, 2026. The executed Axon contract: posted in August 2026; the itemized quote, the Fleet contract and the ACEIP confirmation are not. The Flock audit records: refused Sept 9, 2025 under a records exemption. The announced ALPR dashboards: the Axon portal live; the Flock portal reduced to an overview paragraph. Roney’s 29 written questions of March 2025: no answers published. The quarterly audits: internal, unpublished. (Promise Tracker · Documents)
Cost and savings
The statement. Mayor Manheimer, May 28, 2026: “we received a grant from the federal government to be able to pay for our own,” and described Roney as voting “against a grant that would help save the city money.” (BPR)
The record, from the city’s May 12 staff reports (Documents).
- Before the RTIC: $30,000 a year to the county for Fusus access.
- After: the grant pays for the room and prepays Fusus (about $467,000) for about seven years. The city’s bundled Axon payments are $740,000 in FY27 and $777,500 in FY28 per the staff report. The software renewal after 2033 is at the price then set by Axon.
- The staff report’s documented bundling savings over the 7.5-year contract: $46,948.25.
- Axon’s investor materials report net revenue retention of 124 to 126 percent for more than 20 consecutive quarters; a resident submitted the investor deck as public comment on May 12, and it is in the city’s meeting folder.
- Patrick Conant asked council on the record, “What is the ongoing operational cost after the grant runs out?” No answer has been published as of Sept 23, 2026. The grant and the city’s share, worked through against the staff report and the signed schedule: the federal grant and the city’s share of the contract.
Staffing
The statement. APD’s presentations from March 2025 to May 2026 gave staffing shortages as a reason for the RTIC, which would let the department “do more with less.”
The record. Stepp, April 28, 2026: “there’s no additional staff being requested to support the system today” (the grant and the contract). No staffing cost for the center has been published. The Axon payments are $740,000 in FY27 and $777,500 in FY28. The one randomized trial of Axon’s Draft One report-writing product found no time savings (What Axon Contracts Do). The LAPD Inspector General’s sample found one stolen-vehicle alert in three false (Camera network misuse, documented cases).
The clearance-rate figures, by date
The statements, all to the Public Safety Committee: Chief Lamb, July 2025, violent crime down “37%” year to date. Bo Hess, November 2025: “an over 36% reduction in violent crime.” Interim Chief Stepp, January 2026, the annual figure: “a 24% reduction in violent crime last year.” Lamb, November 2025: “a multi-year 100% clearance rate” for homicides.
The record. The 2025 annual figure is the 24 percent reported in January 2026. At the November 2025 meeting the deputy chief’s figures showed two 2024 homicides still open. The committee minutes record no reconciliation of the figures.
The sheriff’s 2022 crime figures
The statement. Sheriff Miller has credited the county RTIC for consecutive annual crime reductions, including a “15% countywide crime drop” in 2022. (WLOS)
The record, from the SBI annual summaries.
- SBI Table 5 gives the 14.7 percent decrease for the Sheriff’s Office jurisdiction, the unincorporated county, about 28 percent of county crime. Countywide, 2022 index crime fell 5.6 percent and violent crime rose 12.7 percent. Then-Rep. Chuck Edwards made this reading public in a 2023 op-ed titled “No, crime is not dropping in Buncombe County.”
- Countywide rates fell from 2019 to 2020 and from 2020 to 2021, before the county RTIC was operational.
- In 2024 the Sheriff’s Office jurisdiction rose 1 percent while the state fell 4.2 percent; the office’s burglary count rose 42 percent, after a 52 percent rise in violent crime in 2023.
- Countywide index rates from 2021 to 2024: Buncombe County, with an RTIC, down 24.6 percent; Catawba County, with no crime center, down 42.2 percent. Within Buncombe, the largest decreases were in Asheville, which had no RTIC of its own.
- No independent evaluation of the county RTIC has been published. (Justice Quarterly)
When citing these figures, name the jurisdiction (the Sheriff’s Office slice, the county, or the city), the SBI table and the year.
The vendors’ claims
The city’s own Technology and Public Safety page repeats several of the vendors’ marketing lines, and its answer on private cameras reproduces the Axon-built registration site’s text; the two are set side by side on the city’s page, annotated. The six statements Axon and Flock representatives made at the April 28, 2026 work session (the immigration filter, APD’s control of its data, mandatory search reasons, quarterly audits, city data ownership, no facial recognition) are set beside the record in the CAMS and Sunshine Labs one-pager. In the same season, researchers gained root access to a Flock camera in 30 seconds, about 70 cameras were found streaming openly, and members of Congress asked the FTC to investigate (Flock Safety). Vendor statements in other cities that were later contradicted on the record are on Communities that ended or limited camera contracts (Sedona, Oshkosh, Evanston).
Related: Promise Tracker · Camera network misuse, documented cases · Claims Library · Documents