Robotics & AutomationRegulatory PolicyEconomics

The DC Robotaxi Bill Is Really Four Fights at Once

Four fights meet in one permit measure, and Congress can still block it.

Josh LynwoodFounder
Published
Read time
12 min
Share
Where Things Stand

Under the Jobs-Versus-Independence Fight Sits Industrial Policy

For more than seven hours on July 13, 81 witnesses across 17 panels filed through a hearing room in the Wilson Building to tell the DC Council what driverless taxis would do to their city. Teamsters, SEIU 32BJ and transit-union members in the seats called the bill a job killer written for Silicon Valley billionaires; blind and disabled riders called the same technology an independence they had never reliably had; Waymo pledged hundreds of local jobs, and Uber warned that a driverless-only law would hand its rival a monopoly. The measure on the table, the Autonomous Vehicle Deployment Authorization Amendment Act of 2026, would give the District Department of Transportation something it has never run: authority to permit fully driverless commercial service, hedged with seven-figure fees, a per-mile tax, an eight-ward service-equity mandate and a fund for displaced drivers. Under the hearing's jobs-versus-independence binary sits the machinery, DC trying to shape how a safety-critical robotic system enters its streets using the levers of industrial and labor policy, at a moment when the federal government is pressing it to move faster and the technology's own record, gathered in other cities because the District has none, keeps failing in the structured edge cases the crash statistics do not capture. That the guardrails may be aimed at the revenue and liability the bill can price rather than at the failure modes regulators are acting on is our reading of the record below, not a charge we level at anyone.

  • The Bill. Councilmember Charles Allen (Ward 6), who chairs the Committee on Transportation and the Environment, announced the measure on April 23 and formally introduced Bill 26-0684 on May 1 with Brooke Pinto (Ward 2) and Matthew Frumin (Ward 3). It amends DC's 2012 Autonomous Vehicle Act, which today allows testing only with a safety operator behind the wheel, to let DDOT permit fully driverless testing and then commercial passenger and goods service; the Bowser administration has issued no driverless permits, so this is net-new authority, not a tweak.
  • The Hearing. The Committee on Transportation and the Environment held its public hearing on Monday, July 13, at the Wilson Building; it ran more than seven hours and drew 81 witnesses across 17 panels. The Teamsters, SEIU 32BJ, ATU Local 689 and the machinists rallied against the bill and recruited Uber and Lyft drivers to testify, while the American Council of the Blind and Waymo testified in favor; the bill remains in committee, with no markup or Council vote scheduled.
  • The Guardrails. The design is the story: a $1 million application fee plus a $5 million permit against about $5,500 for a human ride-hail license, a 15-cent-per-mile tax, a 200-vehicle cap per company, a 250,000-in-DC-mile testing wall and no commercial service before January 1, 2028. It bolts on an eight-ward comparable-service mandate, a $5 million liability-insurance floor separate from the permit fee, and an Autonomous Vehicle Deployment Fund split evenly between public transit and a workforce fund for displaced rideshare drivers.
  • The Fight. Both sides are real welfare claims on different populations. Teamsters Local 639 president Bill Davis called the bill's so-called guardrails a failure to protect middle-class professional drivers from Big Tech, and SEIU 32BJ's Jaime Contreras called it a job killer; disability advocates testified that robotaxis grant mobility independence; Uber, straddling the field, opposed the driverless-only design and lobbied for a mandate that AVs also carry human drivers, warning the bill would hand Waymo a de facto monopoly; and the DC Chamber of Commerce backed the bill while urging the Council to remove the 200-vehicle cap and lower fees.
  • The Record. DC has no driverless service yet, so the Council is legislating on other cities' incident data, and that record is unsettling in a specific way. Twelve days before the hearing, NHTSA called driverless interference with first responders a functional insufficiency and said it would schedule meetings with developers by month's end to hear fixes; a Waymo struck a 9-year-old in a Santa Monica school zone; the NTSB found a Waymo passed a stopped Austin school bus after its remote assistant answered no to the car's school-bus query; and empty Waymos circled an Atlanta cul-de-sac at dawn, eight of them getting stuck.
Sources13See all 48
Is DC building a framework for the technology's competence or for its failures?

The Guardrails Price Risk Rather Than Validate Safety

Bill 26-0684 would let DDOT issue three-year, revocable permits for driverless commercial service, but only after a wall of conditions: a $1 million application fee and a $5 million permit against about $5,500 for a human ride-hail license, a 250,000-in-DC-mile testing certification, a 200-vehicle cap per company, a 15-cent-per-mile tax, an eight-ward equity mandate, a $5 million insurance floor and no on-demand service before January 1, 2028. It also deems the automated-driving-system maker the legal operator whenever the system is engaged, with a rebuttable presumption if the logs show it active within 30 seconds of a crash, a rule specialists call a national outlier.

The record the Council is weighing comes from other cities, because the District has none of its own, and it says two things at once. Waymo reports about 94 percent fewer serious-injury crashes across 220.6 million rider-only miles, peer-reviewed at a smaller mileage. The failures regulators are acting on are a different animal. Twelve days before the hearing, NHTSA called driverless interference with first responders a functional insufficiency; a Waymo struck a 9-year-old in a Santa Monica school zone; and in Austin the NTSB found that a remote assistant answered no when a car asked whether it had stopped for a school bus with active signals, and the car passed the bus.

The judgment here, a reading of the bill text rather than a verdict on the vehicles, is that the safety layer is real and stops short of what that record says matters most. The bill requires a first-responder-interaction plan, a continuity-of-operations plan and quarterly crash reports the city must publish. It does not require independent validation of the safety case, published The Operational Design Domain, the mapped, geofenced conditions a driverless system is designed to handle, such as certain streets, speeds and weather., or a school-zone performance standard. The center of regulatory gravity has moved from crash rate, where these vehicles look best, to edge-case behavior, where the documented failures cluster. Insurance, caps, a tax and a self-reported plan are calibrated to what a city can price.

Sources28See all 48
Intersections

DC is writing the country's strictest AV rulebook into a federal squeeze

Regulatory Policy. Bill 26-0684 hands DDOT a permit regime it has never run and prices entry in the millions, putting DC at the restrictive pole of a state-by-state patchwork that runs from California's two-agency DMV-and-CPUC model to Texas's preemptive light touch. The live variable is federal: the SELF DRIVE Act and NHTSA's drive to prevent a patchwork target exactly this kind of high-friction local design.

Public Administration. The novelty is administrative, an eight-ward equity mandate, a company-funded workforce fund and phased vehicle caps that turn a yes-or-no on robotaxis into local industrial policy, but the binding constraint is capacity: the AV study the Council leaned on arrived roughly four and a half years late, and DDOT must now stand up permit adjudication, a per-mile-tax system and crash-data intake before the 2028 gate.

What would make this wrong

An enacted act clears the 30-legislative-day congressional review with its per-mile tax, fees and vehicle cap intact, and no appropriations rider follows. The federal squeeze never bound.

Open question

On the federal side: the fees, caps and per-mile tax are the high-friction design a national anti-patchwork push targets, so which of them survives Home Rule review?

Sources4See all 48
The Weave

The Weave maps a single development across domains and across time. Each row follows one domain from where things stand now through the next eighteen months, and expands for the reasoning behind that trajectory.

Wiiver
SECTOR / DOMAINclick a domain to expand
As It Standsthe current status
Immediate0–6 months
Near-Term6–18 months
Technology + Engineering
NHTSA has called driverless interference with first responders a functional insufficiency rather than a rare edge case.
The record regulators are acting on
NHTSA's July 8 directive makes first-responder interference a federal priority.
Guardrails versus failure modes
The bill answers documented edge-case failures with insurance, caps and a plan.
Waymo reports about 94 percent fewer serious-injury crashes across 220.6 million rider-only miles inside its mapped domain.
Safe-per-mile is not safe-in-edge-cases
Safe-per-mile in a geofenced ODD is not the same as safe in structured edge cases.
Modular ML, vendor-graded
The deployed stack is modular ML, and its safety case is vendor-self-benchmarked.
Government + Policy
No fully driverless commercial service is legal in DC; the 2012 Act permits an AV only with a licensed operator aboard.
DC's own AV rulebook
New DDOT permit regime makes DC the US patchwork's most restrictive AV tile.
Fees as a moat
Seven-figure fees and a per-mile tax gate entry to the best-capitalized firms.
Business + Markets
Waymo runs an owned fleet on a $16 billion round; Uber lobbies for a mandate that AVs ride a human-driver network.
Surplus up, income down
Riders gain convergent pricing and access; the driver base bears the income loss.
Which structure DC locks in
The permit picks a structure: Waymo owned-fleet vs Uber's network mandate.
wiiver.co · 4 impacted domains shownWiiverv1 · July 15, 2026
Looking Forward

The Markup Decides Whether the Guardrails Are Real

Strip away the hearing's theatrics and the durable question is design: whether DC uses this bill to validate a technology's competence or to price its failures. The bill is in committee with no markup noticed, and the honest counter is that heavy local guardrails could simply push Waymo to a friendlier jurisdiction and leave DC with neither the jobs nor the accessibility gains, a trade a reader should weigh before treating the guardrails as free. What follows are the checkable signals.

  • The Markup. Whether the Council amends the bill before it votes, and specifically whether any independent-validation, operating-domain-limit, vulnerable-road-user-performance or data-transparency requirement is added, is the tell for whether the safety layer graduates from a self-reported plan to a validated standard.
  • The Home Rule Review. Whether Congress lets an enacted act stand. Any DC law faces a 30-legislative-day congressional review and possible disapproval or an appropriations rider, and with House Republicans already pressing for faster deployment, the live risk is federal pressure on the tax, caps and labor guardrails rather than on robotaxis themselves, our forward read.
  • The 2028 Gate. Not a date on a calendar. The January 1, 2028 launch is a bill provision contingent on passage, and it would fall under a mayor not yet elected: Bowser is a lame duck whose term ends in January 2027, so even an enacted bill would hand the actual buildout to her successor, one from an AV-skeptical field of possible candidates. Read it as forecast, never as a scheduled milestone.
  • The Data Question. Whether markup adds any rule on retention, transparency reporting or law-enforcement access to the fleet's sensor data. As drafted the bill is privacy-blank, so absent an amendment the regime defaults to warrant practice and a company disclosure policy, against a Fourth Amendment line that Chatrie is tightening.
  • The Market Lock-in. Whether DC picks a structure or lets one pick itself. An owned-fleet permit favors Waymo's model; an Uber-style mandate to carry human drivers preserves the aggregator layer and some driver pool, and the same hybrid-network push is spreading to other jurisdictions, so DC's choice is early precedent for who the AV era rewards, our labeled forward read.
Sources8See all 48

Every issue

  1. 01Intersection of the week
  2. 02Impact of the week
  3. 03The week in review

Wiiver Weekly

One free email, Saturdays at 7:00 AM ET.

Unsubscribe anytime.

Sources and Verification
18 of the 48 sources cited here are primaryfilings, opinions, statutes and agency releases read directly
Primary sources18
Secondary sources, by sector30
Government + Policy7
Business + Markets11
Technology + Engineering12
v2 · Reviewed by Josh Lynwood · July 15, 2026
Back to top