Four UC Berkeley graduates have launched AV Watch, a platform for reporting autonomous vehicle incidents like reckless driving and accessibility issues, highlighting a data gap left by official crash‑only reporting.

UC Berkeley graduates have rolled out AV Watch, a public reporting platform for autonomous vehicle incidents like reckless driving and accessibility issues — a direct response to what the team calls a "glaring lack" of safety regulation and data collection beyond crashes.

The platform, built as a capstone project by four Master of Information Management and Systems graduates, won the Dr. James R. Chen Award from UC Berkeley's School of Information. It allows riders and bystanders to submit reports with location, photos, and timestamps, aggregating them on a public dashboard alongside external sources like Reddit.

Cap‑table reality: AV Watch is a graduate‑student project, not a funded startup — there’s no Form D filing or venture backing. The founders, Evan Haas (user research and policy) and Noah Baier (backend engineer and data lead), developed it after a survey of their own found that more than half of respondents had a negative experience with an autonomous vehicle, and 86% of those never reported it. None used a government channel.

The gap AV Watch aims to fill is documented: California’s DMV only requires AV operators to report crashes and disengagements, not subjective driving behavior or accessibility failures. Waymo’s fleet, for example, isn’t wheelchair‑accessible; its wheelchair‑accessible option uses human‑driven vehicles, not autonomous ones.

Context: The launch comes as California considers new AV regulations (AB‑2193 would assign traffic citations to AV manufacturers; SB‑1246 would penalize companies for vehicles that obstruct traffic) and amid heavy lobbying — Waymo made 678 lobbying contacts with San Francisco officials in 2025, the most of any company.

What’s still unconfirmed: whether AV operators will engage with the crowdsourced data, or whether the platform can scale beyond its current proof‑of‑concept stage. The dashboard is live, but the team hasn’t released summary statistics from the reports collected so far.