Summary

Protect drivers before robotaxi service starts or grows

Where the law permits, city and state officials should require robotaxi companies to secure funding for two protections before officials authorize service, renew an authorization, or permit expansion: protection against robotaxi-related losses for covered drivers, and a separate benefit for drivers who qualify through verified past work. Drivers whose work may change should help shape the complete proposal and decide whether to support it.

Public rules and driver protection

Robotaxis could make travel safer and give people who cannot drive greater independence and more transportation options. People should be able to welcome useful technology with confidence that working people will be protected. Protecting people's lives and livelihoods is a basic responsibility of government. Rideshare drivers helped build—and still sustain—the app-based passenger market robotaxi companies seek to serve. Alphabet, Google's parent company, owns Waymo; Amazon owns Zoox; and Tesla reported operating its own robotaxi service. If robotaxis take trips human drivers otherwise would have made, those companies gain revenue while drivers lose work and income. Potential benefits for riders do not justify leaving drivers and their families to bear those losses.

Where law permits, we recommend that public officials adopt enforceable rules requiring robotaxi companies to secure funding for both worker requirements before officials authorize service, renew an authorization, or permit expansion: first, protection against robotaxi-related losses for the driver workforce covered by the proposal; and second, a separate, meaningful additional benefit for drivers who qualify through verified past work—one that should improve their security, opportunity, or freedom over their time. Loss protection responds to harm; the additional benefit recognizes the value drivers helped create.

Because robotaxis use public streets, whether and how they operate is a public choice where law gives officials authority. Those rules should require robotaxi service to benefit the public. Officials should evaluate how robotaxi service, including commercial pilots, affects drivers' earnings from driving and the amount of paid driving work available. Those effects should be core measures of success alongside safety, accessibility, and transportation outcomes. Once service operates, officials should also check whether the required loss protection and separate additional benefit are delivered. Drivers whose work may change should help shape and evaluate the rules before adoption. Public officials remain responsible for setting and enforcing the rules and using independent evidence to decide whether service may begin, continue, or expand.

Here, robotaxi service means paid passenger service in which an automated system does all the driving within defined conditions. Our research focused on people who earned income through Uber, Lyft, or both; some also had experience in taxi, black-car, or local car-service work. Workers First AI neither assumes robotaxi service is inevitable nor recommends stopping it everywhere. Depending on local law, officials may authorize service, authorize it with limits, keep existing restrictions, or refuse authorization.

What participating drivers told us

Thirty-three of the 36 survey respondents said driving was their only or main income. Asked how quickly they could find similar income if they had to stop driving, 23 said they did not think they could, and 10 more expected it would take more than six months. Respondents could select up to two listed forms of support. Twenty selected driver ownership or a share of autonomous-service revenue, 14 selected a one-time payment, eight selected temporary monthly payments, and four selected training and help finding new work.

We interviewed rideshare drivers in New York City, San Francisco, and Atlanta, held recorded group discussions, received 36 survey responses, and drew on unrecorded conversations. Most structured interviews began with drivers' work; interviewers introduced robotaxis and policy options later. The survey listed possible forms of support. The interviews, group discussions, and survey responses describe participating drivers, not the full workforce, and do not support representative estimates.

Participants took three broad positions. Some opposed robotaxi service or supported existing restrictions. Others expected human-driven and autonomous service to coexist under limits. A third group expected expansion and wanted to use drivers' remaining influence to secure protection and a share of the gains while officials decide whether and how service expands. Expecting expansion did not mean supporting it.

Four priorities recurred:

  1. Earning through driving.
  2. Avoiding lower pay or less paid work.
  3. Recognition for past work and the value drivers helped create.
  4. A say in what driver support should include.

Some drivers discussed cash, but not as a complete answer. One Atlanta participant said, “Some of us might be happy to … get the basic income … but we want more. We want to be known that we're valued.” In this report, cash that replaces lost work or income is loss protection; cash may also be one form of the separate additional benefit. The evidence does not establish cash—or any other single form—as drivers' collective preference.

Published research, submitted proposals, financial modeling, advisor feedback, outside expertise, and drivers' later evaluation of policy ideas informed the rules. The project team is responsible for the synthesis and recommendations; participation or review does not imply endorsement.

Evidence and uncertainty

Through September 23, 2026, we identified no rigorous published U.S. study isolating robotaxi service's effects on drivers' earnings, trips, or hours.

Gridwise, Q4 2024 to Q4 2025
Found mixed trends: trips per hour fell in all five markets with robotaxi activity and nationwide, while hourly gross pay excluding tips rose in two of the five markets and fell in three. The comparisons do not show whether robotaxis caused those changes.
Uber, 2026
Reported declines in hourly earnings and in driver utilization in San Francisco and Los Angeles, without isolating robotaxis from other local changes.
Wuhan, 2026 study
Estimated a 10.9 percent short-run decline in average daily operating income per taxi in the district where robotaxi service began. It does not measure individual net earnings or establish what would happen in a U.S. passenger market.

The available evidence therefore does not establish whether or how much robotaxis have changed drivers' work and income. That uncertainty is why public officials should require initial worker protections and secured funding before service begins or expands. Local rules should state how officials will use independent analysis of company records to decide whether to change required company funding, the payments and benefits available to drivers, the amount of robotaxi service allowed, or the conditions under which it may operate.

Recommendations in brief

Key terms

A local proposal should distinguish four overlapping groups of drivers:

  • Covered drivers: the driver workforce within the proposal's scope.
  • Drivers eligible for loss protection: covered drivers who qualify for individual loss-protection payments.
  • Drivers eligible for the additional benefit: drivers who qualify for the additional benefit through verified past work.
  • Affected drivers: drivers whose work or livelihood the proposal may change.

Independent analysts and administrators are independent of every robotaxi company the proposal covers. Scope of service means whether, how much, and under what conditions robotaxi service may operate.

The two worker requirements

We recommend requiring robotaxi companies to meet two separate obligations: loss protection and a separate additional benefit. One cannot substitute for the other.

  • Measuring the loss: First, independent analysts must estimate total net robotaxi-related loss across all covered drivers. That includes lost paid work and income, plus any other losses the local proposal expressly covers. The estimate first accounts for changes that would have happened anyway, so only robotaxi-related loss counts.
  • Paying before service: Before service is authorized or expanded, public officials should require robotaxi companies to meet two enforceable funding requirements: one sufficient to cover the best-supported estimate of that loss, and a separate requirement for the additional benefit.
  • Three time periods: Local rules should set how long companies must fund loss protection across covered drivers, how long each individual driver receives payments, and how long the additional benefit lasts. The company's obligation can outlast any single driver's payments. If robotaxi-related loss among covered drivers continues after individual payments end, companies must keep covering it.
  • If money runs short: Loss protection is paid first, and the additional benefit is still required.

What the requirements do not do: They do not guarantee that every driver keeps the same hours or income. They also do not require drivers to prove that robotaxis caused each lost trip or dollar. Training, promised but unguaranteed jobs, and wages from other work cannot by themselves satisfy either obligation. And drivers need fair pay and enforceable rights in human-driven work whether or not robotaxis ever operate.

Five policy areas

  1. Evidence about drivers' work and income — require company records, independent analysis, published methods, and privacy protection, plus rules stating how officials will use new evidence to determine whether required company funding, the support available to drivers, or the amount or conditions of robotaxi service permitted should change.
  2. Funding — secure company funding before service begins or expands, with published amounts, payment methods, correction deadlines, and rules for what happens if funding falls short.
  3. Scope of service — depending on local law, preserve restrictions, decline authorization, require that a defined share of passenger service remain human-driven, or limit vehicles, trips, hours, areas, or expansion. Officials should not automatically renew an authorization or allow more service.
  4. Loss protection and a separate additional benefit — base company funding on losses across all covered drivers. If evidence points to serious losses in a particular group, type of driving work, or location that the overall estimate masks, independent analysts should review the evidence, and officials should decide what protective action is needed before further expansion.
  5. Eligibility, division of payments and benefits, driver decisions, and administration — define who is covered, who qualifies for each payment or benefit, how payments and benefits are divided, how the process determines whether affected drivers support the complete proposal, how appeals work, how records and personal information are protected, who administers the rules, how company control is prevented, and what must be reported publicly.

The report also compares approaches that address different questions, including continued restrictions or the decision not to authorize service, limits that preserve human-driven work, license-and-royalty arrangements, debt relief, a fund that invests transition revenue before making payments, pooled driver ownership, trusts that hold company shares for workers, collective bargaining, and cooperatives. The project did not adopt any submitted proposal in full. The project still recommends the two worker requirements. Affected drivers should help decide which approaches a local process develops.

Two driver decisions

The proposal requires two separate driver decisions; neither can replace the other.

  • Affected drivers should shape loss-protection rules. A process using a published rule for measuring support must establish their support for the complete local proposal.
  • Drivers eligible for the additional benefit select or approve the kind of benefit and how it is divided.

Both decisions require paid, accessible, multilingual participation without company control. Rules must say how any representatives are chosen and held accountable. Non-participation is not support. Significant changes require affected drivers to decide again; changes to the benefit or how it is divided also require drivers eligible for the additional benefit to approve the revision. Officials adopt and enforce the rules; independent administrators verify records, calculate payments, and apply the rules.

Driver support does not authorize service or waive worker, company, or public duties. Officials retain their separate legal and authorization decisions and may deny authorization even when both driver decisions succeed.

Four questions before authorizing robotaxi service

  1. Legal authority. Do the relevant public bodies have the legal power and practical ability to enact, administer, and enforce the proposal, including the power to require the specified company funding?
  2. Financial feasibility. Does independent review find that, at the proposed fare, amount of service, and service design, the companies responsible for robotaxi service can finance both worker requirements and every other requirement the proposal places on them? Separately, does every other public requirement that requires funding have an adequate funding source?
  3. Driver decisions. Through paid, accessible, multilingual participation free from company control and under rules published in advance, did affected drivers support the complete proposal, and did drivers eligible for the additional benefit select or approve the benefit and how it is divided?
  4. Other public requirements. Does the proposal meet applicable safety, accessibility, transportation, privacy, consumer-protection, civil-rights, environmental, and other public requirements?

Our illustrative model estimates how much selected charges on robotaxi fares could raise before costs under stated assumptions; it does not establish driver losses, costs, profitability, or local feasibility.

Where law permits, public officials should not authorize new or expanded service if independent review finds that the companies responsible for robotaxi service cannot finance both worker requirements at the proposed amount of service. Responses may include more company funding, less service, continued restrictions, or a revised proposal. Public money may add support or finance a specific public purpose the proposal identifies and justifies—such as transit, infrastructure, or accessibility—but should not reduce required company funding or excuse any other requirement. Local law determines available actions.

Possible next steps

Further work should proceed only where affected drivers and their organizations want to participate. A local effort may oppose service, preserve restrictions, negotiate conditions, compare alternatives, or develop one complete proposal for drivers to evaluate. Claims of public benefit require evidence. Starting or expanding service is not the measure of success.

This report and its method are a starting point—not jurisdiction-specific legal advice or a completed local financial-feasibility model. A local effort based on this report would require a separately defined effort, willing local partners, appropriate expertise, and resources.

Technologies, institutions, laws, workforces, and remedies differ across sectors, so these policies do not transfer unchanged. Working people in each sector should decide what protects or improves their lives and how to share in automation's gains. Broader approaches require separate research and public decisions.

Conclusion

Where applicable law gives public officials authority, whether commercial robotaxi service may operate—and under what conditions—is a public decision, not a force of nature. That makes the central question public: if robotaxi service comes to a city, who bears its costs, and who shares in its gains? Drivers and riders should be able to help shape the rules before public officials make that decision.

The five policy areas identify what a local proposal must address to make both worker requirements enforceable. Evidence alone does not secure the funding needed to protect drivers against losses. Funding without service limits may let expansion outpace the evidence needed to show whether the funding remains sufficient. Limits without loss protection leave existing harm unresolved. The complete proposal is not legitimate unless affected drivers support it and drivers eligible for the additional benefit decide what the benefit will be and how it will be divided. Public officials still decide whether to authorize robotaxi service.

Participating drivers did not speak with one voice. Some asked that robotaxi service not come at all; others asked for limits, study before expansion, or a share of the gains. Across those differences, recurring priorities were to keep earning, protect pay and work, be recognized for what drivers helped build, and help decide what comes next.

As a New York driver put it, “But what I want to highlight is that before taking someone's livelihood, give them something.”

Success is not measured by whether robotaxis arrive. It is measured by whether the people who helped build this market are protected and share in what comes next. In the Atlanta discussion, one participant asked only this: “Just do not give up on us.”

To follow the project and what comes next, get updates. For questions or interviews, contact the project team.