What Are the Risks?

CVI has documented evidence and broad support, but it also has documented failures, structural vulnerabilities, and significant evidence gaps. What follows presents each risk with its source.

Federal Funding Fragility

In April 2025, the DOJ terminated more than 360 grants valued at $811 million across public safety programs, per Reuters and the Council on Criminal Justice. Baltimore’s ROCA program lost $1 million overnight, per Baltimore Sun reporting. LifeBridge Health, which operates six Safe Streets sites, lost $1.2 million, per Baltimore Banner reporting. Detroit’s FORCE program lost approximately $2 million and laid off frontline workers, per Bridge Michigan. Memphis HEAL 901 had its $1.7 million grant canceled after receiving only $150,000, per Stateline — the executive director dipped into personal savings to keep 14 staff members on payroll.

The class-action lawsuit was dismissed in July 2025 but is on appeal in the D.C. Circuit. The FY2026 federal budget proposes eliminating CVI grant funding entirely, per Giffords.

Angela Kimball of Inseparable, a national mental health and public safety advocacy organization, described the structural mismatch: cities pay police and fire for the capacity to be ready — officers are funded whether or not they respond to a call. CVI is funded for active response — Medicaid reimburses billable encounters. The unfunded gap includes training, supervision, travel time, community relationship-building, and staff availability. That mismatch, not program performance, explains most collapses. Dr. Kyle Fischer described Medicaid reimbursement as providing “stability” but cautioned it is “not a panacea.”

Workforce Safety

CVI workers operate without badges, weapons, or legal authority in their cities’ highest-violence neighborhoods. The 2023 ProPublica investigation documented an interrupter who physically accompanied a threatened mother on the bus for several days and another who talked down a man holding two guns. In Minneapolis, Sasha Cotton reported in 2022 that interrupters mediated more than 1,500 incidents before they became violent.

No systematic catalog of CVI worker injuries or deaths has been published. The occupational risk is inherent to the model.

Burnout and secondary trauma are documented occupational hazards, . Workers maintain relationships with people who are killed despite the intervention. They attend funerals for clients. They carry the weight of failures alongside invisible successes. Workforce retention depends on organizational infrastructure — supervision, trauma support, competitive compensation — that many programs have not adequately built. The emotional toll of the work and competition from better-paying positions compound the retention challenge.

Coverage Gap

Chicago’s CRED currently serves approximately 15% of the individuals assessed as highest-risk, with a goal of 75%, per the SC2 initiative and the Chicago Tribune. That gap between identifiable need and actual capacity exists in every documented city.

The workforce pipeline constrains scale: the University of Chicago’s Leadership Academy takes six months per cohort and has reached 21 cities, per the Chicago Defender, but the credible messenger credential — lived experience with violence — cannot be mass-produced.

Evidence Limitations

Three RCTs have evaluated CVI or CVI-adjacent programs: READI (2,456 participants, QJE 2024), Denver GRID (143 participants, NIJ 2024), and Choose to Change (2,074 youth, UChicago Crime Lab). All found reductions in violence. The generalizability of these results to different programs in different cities remains an assumption, not a finding.

An important statistical note on READI: the pre-specified primary outcome did not reach conventional statistical significance (p=0.13 after multiple-testing adjustment). An independent review at EvidenceBasedPolicy.org characterizes the effects as “suggestive under established scientific standards.” The cost-benefit ratio (4:1 to 18:1) is statistically significant (p=.03).

A February 2026 systematic review of 83 Cure Violence evaluations, published in INQUIRY, found that 68.7% showed violence reductions but only 32.5% reached statistical significance. The majority of CVI programs report their own outcomes without independent verification. City-level claims like Detroit’s 72% decline in targeted neighborhoods, Philadelphia’s 26% homicide reduction, and Lansing’s 52% fatal shooting decline are program-reported or city-reported data, not independently evaluated findings.

The Lansing evaluation’s own authors flagged that the decline was greater in non-program areas than in program areas, per the Michigan Advance — raising questions about specific program attribution versus citywide or national trends.

A 2025 St. Louis evaluation of the Life Outside of Violence HVIP program, published in the Annals of Internal Medicine, found no significant difference in reinjury — the field’s most prominent null result.

A February 2024 White House estimate credited CVI with contributing to the 12.4% national homicide decline, but this is a modeled attribution across a broader strategy including policing and gun legislation, not a direct measurement of CVI’s isolated effect.

Long-Term Data

Johns Hopkins tracked Baltimore Safe Streets from 2007 through 2022 — the field’s only long-term longitudinal evidence. The 2024 READI results show the 44% gunshot victimization decrease sustained at 18 months. Beyond these, long-term outcome data is limited.

Geographic Concentration

The documented programs and evidence base are concentrated in large cities with progressive political environments, strong nonprofit sectors, and existing public health infrastructure: Chicago, Baltimore, Detroit, Orlando, New York, Philadelphia. Pine Bluff, Arkansas and Fairbanks, Alaska represent small-community exceptions with limited published data, .

Whether the model produces equivalent results in politically conservative jurisdictions, rural communities, or different institutional landscapes is not documented.

CVI-Police Information Sharing

Data Currency Risk

Every statistic in the CVI field carries a date, and conditions change. The April 2025 DOJ contraction eliminated programs that were producing documented results at the time of their termination — FORCE Detroit’s Keepers program “had been credited with helping reduce homicides to the city’s lowest since 1965,” per Bridge Michigan, when it lost approximately $2 million. Fresno’s Advance Peace “had seen homicides decline 59% since the program launched,” per CalMatters, when it lost $2 million. Oakland’s Youth ALIVE!, operating what multiple sources describe as the nation’s first hospital-based violence intervention program, lost $2 million.

The Community-Based Public Safety Collective, which was providing technical assistance to 95 federal grantees — the national infrastructure for new CVI program implementation — lost its cooperative agreement entirely, per the Brennan Center. CDC layoffs in April 2025 “decimated” the Injury Center’s Division of Violence Prevention, per The Trace (Chip Brownlee, April 2025, https://www.thetrace.org/2025/04/cdc-layoffs-gun-violence-prevention-trump/) — eliminating some of the federal research capacity that supports CVI evaluation.

The compound effect of ARPA expiration ($2 billion, deadline December 31, 2026), DOJ grant terminations ($811 million), and the proposed FY2026 elimination creates compound pressure. Programs that built sustainability plans around replacing one federal source now face the loss of multiple sources simultaneously.

For a decision-maker evaluating CVI, the risk landscape reduces to a documented question: does the program have the funding diversification, research partnerships, and institutional embedding to survive the current federal contraction? The programs that answer yes — those in the nine Medicaid states (per HAVI), those with state appropriations, those with local budget lines and philanthropic backing — have documented a survival path. The programs that answer no face the same structural vulnerability that collapsed ROCA’s $1 million overnight (per Baltimore Sun) and forced Memphis HEAL 901’s director to personal savings (per Stateline).

The 2023 ProPublica investigation described the tension: CVI workers build trust by operating independently from law enforcement. Police investigators need information CVI workers may have. If a credible messenger is perceived as a police informant, their access — and their safety — is compromised.

The operational protocols governing this tension are not well documented in public sources. The ProPublica investigation described the challenge as largely unresolved.

Rural and Suburban Applicability

The documented CVI programs and evidence base are concentrated in high-violence urban areas. Pine Bluff, Arkansas (population roughly 40,000) went 543 days without a juvenile homicide, per THV11, and Fairbanks, Alaska operates a program. Anne Arundel County, Maryland expanded from the Baltimore Safe Streets model into a suburban context, per the Baltimore Banner (Rick Hutzell, “If gun violence is a disease, Annapolis is about to get its shot,” September 5, 2023, https://www.thebanner.com/opinion/column/cure-violence-eastport-newtown-bay-ridge-annapolis-torri-gedin-3FNPZP5WJFE4JPOW2XGXDFJN24/). But these remain exceptions. Whether the CVI model — designed for urban network-contagion violence patterns — adapts to rural communities with different violence dynamics, thinner social networks, and greater geographic dispersion is documented only in these limited cases.

A core tension for rural applicability: the credible messenger workforce credential — lived experience with violence in the specific community — depends on a density of social networks that exists in high-violence urban neighborhoods but may not exist in dispersed rural settings. The University of Virginia’s “In Their Own Voices” report found that “holistic and comprehensive community responses [like Community Violence Interrupters] are effectively and creatively lowering the rates of gun violence” across Virginia, per the Virginia Mercury, but the documented implementations remain concentrated in urban and suburban areas.

Mechanism Uncertainty

The READI evaluation tested the combined package — outreach, employment, cognitive behavioral therapy, case management — not individual components. Whether the credible messenger relationship, the employment, or the CBT drives the results has not been tested in isolation, per the QJE publication. “No comparative study has tested different CVI program designs against each other.”

The Invisible Success Problem

As one violence interrupter told ProPublica: “We squashed it. But nobody knows about those kinds of stories.” A shooting prevented leaves no trace in any dataset. The mediations that produce population-level reductions are events that did not occur.

As one violence interrupter told ProPublica: “We squashed it. But nobody knows about those kinds of stories.” The population-level evidence captures what CVI prevented in aggregate (fewer shootings in program areas) but cannot capture the individual mediations and interventions that produced those results. Minneapolis’s 8,900 contacts and 1,500 mediations in six months, per Sasha Cotton in 2022, represent documented activity — but the shootings those mediations prevented are, by definition, events that did not occur and do not appear in outcome datasets.

The measurement challenge has political consequences. A program that prevents 50 shootings in a year has no 50 data points to show a city council. It has a statistical decline in a neighborhood’s shooting rate — which could be attributed to CVI, to policing changes, to economic shifts, or to demographic trends. The RCTs address this through randomization, but most programs operate without experimental controls and face the attribution problem directly in every budget hearing.

Bottom Line

CVI’s risks are documented and specific. Federal funding contraction has already eliminated programs and the FY2026 budget proposes complete elimination. The capacity-vs.-encounter funding mismatch described by Kimball and Fischer makes programs structurally fragile. Chicago’s CRED serves 15% of its target population. The evidence base includes three RCTs and one 15-year longitudinal study, but READI’s primary outcome did not reach conventional significance, most programs lack independent evaluation, and the St. Louis HVIP evaluation produced a null result. The CVI-police information sharing tension remains under-documented. Rural applicability is untested. A leader who understands these risks is better positioned to design a program that survives them.

Source Appendix

DOJ terminations — 360+ grants, $811M, city-level impacts. Reuters, April 2025. Council on Criminal Justice, 2025. NPR, November 2025. Stateline, May 2025. Brennan Center, 2025. Baltimore Sun, Baltimore Banner, WYPR.
Lawsuit dismissed, D.C. Circuit appeal, FY2026 elimination. Law360, July 2025. Giffords, September 2025.
Kimball / Inseparable — capacity vs. encounter mismatch. NPR (All Things Considered), Aaron Bolton, “Mobile crisis teams shut down amid funding troubles,” January 1, 2026. https://www.npr.org/2026/01/01/nx-s1-5652076/mobile-crisis-teams-shut-down-amid-funding-troubles Kimball quote: “You need to pay for the capacity to be at the ready, just like we do for fire and police, regardless of whether somebody is going to be called out.” (Note: Kimball’s quote was in context of mobile crisis funding, not CVI specifically; the structural argument applies across alternative response programs.)
Fischer — “stability” and “not a panacea.” Health Affairs Forefront (Fischer, Morris, Piening). https://www.healthaffairs.org/content/forefront/expand-medicaid-s-investment-community-violence-intervention Also: The Guardian, Sonya Singh, June 7, 2023. https://www.theguardian.com/us-news/2023/jun/07/gun-violence-medicaid-california-intervention-program
ProPublica — worker risk, bus accompaniment, information-sharing tension. Alec MacGillis, May 2023. https://www.propublica.org/article/are-community-violence-interruption-programs-effective
SC2 initiative — 15% coverage, 75% goal. https://www.scalecvichicago.org/ Also: Chicago Tribune.
Leadership Academy — 21 cities, six months. Chicago Defender, September 2023.
READI — 2,456 participants, p=.13 primary outcome, 4:1-18:1 (p=.03). Heller et al., QJE, 2024. EvidenceBasedPolicy.org review.
Cure Violence review — 68.7% reductions, 32.5% significant. INQUIRY, February 2026.
Lansing — authors’ caveat on non-program areas. Michigan Advance, June 2025.
Life Outside of Violence — null result. Annals of Internal Medicine, 2025.
12.4% federal estimate — modeled attribution. White House, February 2024.
JHU — 2007-2022 longitudinal. Johns Hopkins Bloomberg School of Public Health.
Sasha Cotton — 1,500 mediations. KSTP, May 2022.
Violence interrupter — “We squashed it.” ProPublica, May 2023.
Detroit 72%, Philadelphia 26%, Lansing 52% as Tier 2 data. See Q05 Source Appendix.
Pine Bluff, Fairbanks — small-community exceptions. THV11, May 2025 (Pine Bluff). https://www.thv11.com/article/news/local/pine-bluffs-gun-violence-intervention-program-500-days-no-juvenile-homicide/91-371d5b84-ded6-438d-b0b7-8a2df3cec869 (Fairbanks: no independent external source located for this claim.)