Executive summary

Waymo’s public response to Senator Markey’s concerns reasserted that remote-assistance agents never assume primary control of vehicles. Yet by stopping short of publishing intervention-rate, latency or audit-log metrics, the company leaves a transparency gap that governance requires. This absence heightens the odds of prescriptive oversight and shapes how regulators, partners and investors appraise the safety assurances of autonomous fleets.

Key takeaways

  • Thesis insight: Waymo’s detailed PR clarification reinforces role boundaries for remote-assist (RA) and U.S.-based Event Response Teams (ERTs) but omits the measurable safety signals regulators and enterprise partners will seek.
  • Operational framing: RA is advisory and event-driven—compared by Waymo to air-traffic control—while certified ERTs handle collisions, towing, law-enforcement interfaces and reporting.
  • Transparency gap: No published intervention-rate per vehicle-mile or ride, no average time-to-assist benchmarks, and no audit logs that would demonstrate how often RA guidance alters ADS decisions.
  • Regulatory risk: Past AV reporting precedents—such as NHTSA’s post-crash data mandates and California DMV disengagement logs—suggest lawmakers could impose standardized disclosure if voluntary metrics remain absent.
  • Market trust: Firms willing to furnish auditable metrics—intervention-rate, response-latency service level objectives (SLOs) and documented ERT handoffs—are positioned to lead industry transparency norms.

Breaking down Waymo’s clarification

In a February 17, 2026 letter to Senator Ed Markey and a concurrent blog post by Ryan McNamara, head of global operations, Waymo laid out three core assertions:

  1. The autonomous driving system (ADS) remains in full control of vehicle functions at all times.
  2. Remote-assist (RA) agents respond only to system-initiated, event-driven queries—never to open-ended requests—and do not take direct control of steering, braking or navigation.
  3. Event Response Teams (ERTs), exclusively based in the U.S., are certified to manage collisions, towing, law-enforcement liaison and regulatory reporting.

Waymo quantified its RA operation as roughly 70 agents on duty worldwide—split between Arizona, Michigan and two Philippine cities—supporting a 3,000-vehicle fleet that logs over 4 million miles and 400,000 rides per week. The company used an “air-traffic control” analogy: RA agents supply high-level context that the ADS evaluates autonomously.

This narrative sought to dispel public confusion triggered by Senator Markey’s labeling of Philippine-based operators as “completely unacceptable,” citing latency and national-security concerns. By framing RA as purely advisory and ERTs as the sole incident handlers, Waymo aimed to quarantine potential liability and cybersecurity risks.

What remains unsaid

Waymo did not publish any of the following metrics:

  • Intervention-rate per 1,000 vehicle-miles or per 10,000 rides, indicating how often RA guidance affects ADS behavior.
  • Average time-to-assist benchmarks—particularly from overseas centers—showing latency performance under load.
  • Audit-log data or recordings that trace the sequence of system-requested RA interventions and their outcomes.
  • Count of incidents escalated from RA to ERT, illuminating the threshold between advisory support and certified response.

Without these figures, Waymo’s rebuttal replaces one form of opacity (whether RA exists) with another (how often and how quickly RA matters). Stakeholders—from NHTSA to enterprise customers—are likely to regard this omission as a signal that voluntary transparency has limits.

Competitive and regulatory background

Remote assistance is not unique to Waymo. Since 2024, Zoox and Cruise have described analogous “fleet response” frameworks in blog posts and safety reports. Tesla’s shadow-mode architecture likewise aggregates human-reviewed scenarios into training data. Yet none of these players has fully disclosed the granular metrics that independent analysts say are critical to governance.

The regulatory environment has grown more exacting. In 2021, NHTSA issued preliminary AV crash data requirements, leading to quarterly incident reports. California DMV’s AV testing logs mandate disengagement counts per 1,000 miles. Congress has held hearings—like Peña’s testimony—with bipartisan calls for standardized reporting. When voluntary disclosures fall short, precedents suggest regulators will default to prescriptive rules: data-residency mandates, mandatory SLA-style thresholds for intervention latency, or formal audit-rights for independent overseers.

Public sensitivity to offshore staffing has intensified under executive orders on critical-infrastructure resilience. Supply-chain scrutiny in other sectors—semiconductors, health data—has already produced restrictions on foreign data processing. Similar pressures could see states require remote-assist centers to reside within U.S. borders or impose background-check regimes on overseas personnel.

Risks to operators, partners and regulators

  • Regulatory risk
    • Federal agencies and state lawmakers could cite the absence of auditable metrics to justify formal rulemaking on RA operations.
    • Industry lobbies may fragment if prescriptive rules differ by jurisdiction, slowing nationwide AV deployment.
  • Security risk
    • Overseas RA raises plausible supply-chain vulnerabilities—cyber intrusions, data-soak attacks and unauthorized access to live feeds.
    • Background-check gaps and cross-border data flows could trigger state-level data-protection statutes or executive actions on critical-tech safeguarding.
  • Reputational and legal risk
    • Confusion over “remote drivers” may fuel litigation alleging deceptive safety claims or local ordinance rollbacks that stall service areas.
    • Investors and enterprise buyers could interpret nondisclosure as a warning flag, driving down private valuations or contract terms.

Disclosure expectations

To close this transparency gap, stakeholders will likely seek a concise set of auditable metrics rather than open-ended reports. Those metrics include:

  • Intervention-rate per 1,000 vehicle-miles and per 10,000 rides, disaggregated by RA versus ERT involvement.
  • Average and percentile latency for RA responses, with separate figures for domestic and offshore centers.
  • Counts of system-requested interventions handled at RA level versus escalations to ERT.
  • Audit-log summaries showing time-stamped decision requests, agent acknowledgements and ADS override outcomes.
  • Data-residency declarations specifying where recorded feeds and logs are stored and who has access.

These are the kinds of disclosures that regulatory bodies, large fleet customers and institutional investors will look for to validate safety claims and manage compliance risk.

Likely stakeholder responses

  • Regulators are likely to propose rulemaking that codifies standardized reporting fields—mirroring NHTSA’s approach to crash data—and to hold follow-up hearings if voluntary metrics remain absent.
  • Enterprise partners may insist on detailed SLAs specifying maximum response-latency SLOs, data-residency conditions and audit-access provisions as conditions for pilot programs or contracts.
  • Investors could factor in potential delays and compliance costs stemming from jurisdictional mandates on remote-assist staffing, adjusting valuation models accordingly.
  • Rival AV operators may align around a transparency framework—co-developed with regulators and standards bodies—to preempt inconsistent state-level rules.

What to watch

  • Whether Waymo or rivals publish the intervention-rate and latency benchmarks in upcoming safety reports or earnings calls.
  • Federal rulemaking initiatives from NHTSA or legislative proposals in Congress targeting RA or data-residency requirements.
  • State-level inquiries—particularly in California, Arizona and Michigan—that could translate into local mandates on offshore staffing.
  • Public statements from Cruise, Zoox and Tesla clarifying their own RA or shadow-mode metrics—potentially establishing de facto industry standards.
  • Independent audits or third-party studies that validate or challenge the anecdotal “air-traffic control” analogy with hard intervention data.

By reframing operational transparency as a core governance signal, the AV industry sets its trajectory for regulatory trust and market acceptance. Absent auditable metrics, companies risk inviting the very prescriptive oversight they seek to avoid.