NYC Autonomous Vehicle Delay Dashboard

Estimated preventable deaths & injuries since Aug 1, 2025
--
Based on peer-reviewed crash-reduction research and a 2.0% effective VMT-share assumption (Kusano et al.)
Preventable deaths
--
85% crash reduction
Total injuries prevented
--
Pedestrian + cyclist + motorist
Pedestrian injuries
--
92% reduction
Cyclist injuries
--
85% reduction
Motorist injuries
--
85% reduction

The Delay

A timeline of NYC's autonomous vehicle regulatory failure

Aug 2025
NYC DOT grants Waymo first AV testing permit
Up to 8 Jaguar I-PACEs in Manhattan and Downtown Brooklyn, safety driver required
Oct 2025
Testing permit extended through end of 2025
Limited testing continues — no public ride-hail service allowed
42 road deaths since delay began
Jan 2026
Governor Hochul proposes robotaxis statewide — except NYC
NYC explicitly excluded from commercial AV operations
104 road deaths since delay began
Feb 2026
Governor withdraws robotaxi proposal entirely
Labor opposition from taxi unions kills even the statewide plan
125 road deaths since delay began
Mar 2026
Testing permit expires
No legal path to commercial AV service exists in NYC
146 road deaths since delay began
Apr 2026
Still blocked — the most restrictive major US city
No legislation, no commercial permit, no driverless operations allowed
169 road deaths since delay began

The Data

Waymo's safety record applied to NYC

NYC Traffic Fatalities (2019–2025)

2019
220
2020
243
2021
273
2022
255
2023
253
2024
231
2025
250*

* 2025: projected from partial year; 2026: projected from YTD

NYC Traffic Injuries (2019–2025)

2019
61,390
2020
44,614
2021
49,653
2022
49,541
2023
51,840
2024
52,194
2025
48,222*
Pedestrian Cyclist Motorist

Waymo Safety Reductions (Peer-Reviewed)

Injury crash reduction -85%
Pedestrian injury reduction -92%
Intersection crash reduction -96%
Miles analyzed 7.1 million

Applied to NYC (annual, based on 2024, 2.0% VMT share)

Deaths prevented/year 4
Pedestrian injuries prevented/year 177
Cyclist injuries prevented/year 88
Motorist injuries prevented/year 636
Total people spared/year 905

2.0% effective VMT share = NYC ride-hail VMT (~9.0%, Open Plans / Streetsblog 2023) × observed Waymo rideshare penetration in SF (~22%).

Source: Kusano et al. (2024) — 85% fewer any-injury-reported crashes over 7.1M autonomous miles

Source: Kusano et al. (2025) — 85% fewer suspected serious injury+ crashes, 79% fewer any-injury-reported crashes over 56.7M miles

Source: Swiss Re / Waymo (2024) — Waymo vehicles had zero bodily injury claims vs. human baseline

Ride-Hail Safety Comparison

Incident rates across ride-hail providers and national baseline

Provider Metric Rate
Waymo Serious injuries / million miles 0.02
Uber Accidents / million miles 0.45
Lyft Accidents / million miles 0.38
National avg. Fatalities / 100M VMT 1.35

Methodologies differ across these sources. Waymo data uses police-reported incidents matched to location baselines; Uber/Lyft figures are self-reported. Direct comparison should be interpreted with caution.

Sources: Kusano et al., Traffic Injury Prevention (2025), 56.7M miles · Uber Safety Report (2021-2022) · Lyft Safety Report (2021-2022) · NHTSA FARS

Meanwhile, Elsewhere

Months from first Waymo testing to commercial ride-hail service

2023
2024
2025
2026
Los Angeles
12 mo
Austin
15 mo
Atlanta
5 mo
Miami
6 mo
New York
9+ mo
Testing → launched Testing → still waiting

Methodology & Sources

Full transparency on data, assumptions, and limitations

How the counter works

The live counter estimates preventable deaths and injuries by interpolating NYC's cumulative traffic casualties since August 1, 2025 (when NYC DOT granted Waymo its first AV testing permit but blocked commercial deployment) and applying Waymo's peer-reviewed safety reduction factors to each category.

NYC crash data comes from the NYPD Motor Vehicle Collisions dataset on NYC Open Data, cross-referenced with Vision Zero reports. Deaths and injuries are interpolated linearly within each year to produce real-time estimates.

Estimate methodology

The headline counter sums four categories, each with its own reduction rate applied to 2.0% of vehicle miles traveled: deaths (85% reduction), pedestrian injuries (92%), cyclist injuries (85%), and motorist injuries (85%). The category-specific rates come from Kusano et al. (2024), which analyzed 7.1M rider-only miles and found 85% fewer any-injury crashes and 92% fewer pedestrian injury crashes. A subsequent 2025 study covering 56.7M miles confirmed 85% fewer suspected serious injury+ crashes (95% CI: 39–99%) and 79% fewer any-injury-reported crashes (95% CI: 71–85%).

The headline estimate uses an illustrative 2.0% effective VMT-share assumption. This is derived by multiplying NYC ride-hail's share of city-core vehicle miles traveled (about 9.0%, Fehr & Peers, 2019) by a benchmark Waymo share of the rideshare market (about 22%, based on San Francisco at end of 2024; data from YipitData).

Key assumptions & limitations

  • The 85% reduction comes from Kusano et al.'s analysis of 7.1 million autonomous miles. Real-world deployment at scale could differ.
  • Fatality reductions are extrapolated from injury crash data. Fatality-specific reductions may be higher or lower than the overall injury crash reduction.
  • The 2.0% effective VMT share is derived from NYC ride-hail VMT (~9.0%, Open Plans / Streetsblog 2023) multiplied by observed Waymo rideshare penetration in San Francisco (~22% by end of 2024, YipitData).
  • These figures represent potential lives saved, not certainties. AV technology continues to improve, and real-world results will depend on deployment specifics.
  • NYC's road conditions, pedestrian density, and traffic patterns differ significantly from cities where Waymo data was collected (primarily Phoenix and San Francisco). NYC's high pedestrian volume could affect outcomes in either direction.
  • NYC ride-hail VMT share may differ from the 2019 Fehr & Peers estimate. Post-pandemic ride-hail usage patterns have shifted in many cities.

Data sources

Disclaimer: This dashboard presents estimates based on publicly available data and peer-reviewed research. It is intended to illustrate the potential human cost of regulatory delay, not to predict exact outcomes. The authors are advocates for autonomous vehicle deployment and present this data in good faith to inform public discourse. All sources are linked for independent verification.