On April 28, 2026, the California DMV adopted what it calls the most comprehensive autonomous-vehicle rules in the United States. The headline is not a new ban or a new permit class — it is data. Operators must now report system failures, vehicle immobilizations and hard-braking events, file expanded collision reports, and disclose vehicle-miles-traveled figures. Police can issue a Notice of AV Noncompliance to a manufacturer when a driverless car commits a moving violation, and local emergency officials can send an electronic geofencing directive that compels a fleet to clear an area within two minutes.
The substance of the California rule is a transparency architecture. For India — where the topic of autonomous-vehicle data policy remains almost entirely unwritten — it is a useful mirror, and an uncomfortable one.
India is mandating the car, not the data
India's regulatory energy is going into hardware, not reporting. From April 1, 2026, the Ministry of Road Transport and Highways requires a defined package of Advanced Driver Assistance Systems on new models of buses and trucks (categories M2, M3, N2 and N3), extending to existing production models from October 1, 2026. The mandate stacks five Automotive Industry Standards — emergency braking (AIS-162), drowsiness warning (AIS-184), blind-spot information (AIS-186), moving-off information (AIS-187) and lane-departure warning (AIS-188).
These are real safety systems, and on a road network that loses over 150,000 lives a year they are defensible. But notice what the mandate does not include: any obligation to report how often those systems fail, how many kilometres they cover, or what they record before a crash. India's separate Event Data Recorder standard, AIS-192, aligned to UN Regulation 160, is still in draft. The 1988 Motor Vehicles Act — the statute under which all of this sits — was written before the word "autonomous" had a regulatory meaning at all.
The governance layer exists, but points at liability, not transparency
India is not a policy vacuum. On November 5, 2025, the Ministry of Electronics and IT released the India AI Governance Guidelines, which reshape accountability so that at SAE Level 3 and above, liability shifts from the human driver toward manufacturers, software providers and OEMs. The guidelines invoke "understandable by design" principles and India-specific bias testing for pedestrian detection.
That is the right instinct on accountability. But it answers a different question than California's. India's framework is increasingly clear about who pays after a failure; it is silent on what gets measured so failures can be seen, compared and pre-empted. Liability without standardized telemetry means the country learns about autonomous-system performance one lawsuit at a time.
The case for a data mandate — stated fairly
The strongest argument against importing California's regime is that India should not regulate a market it barely has: there are no robotaxis on Indian streets, and a heavy reporting burden could deter the ADAS and logistics-autonomy investment that India is well-positioned to win. That case has force, and a precautionary instinct runs deep here — in 2017 the then-transport minister said India would not allow driverless cars at all, to protect drivers' jobs.
But California's design rebuts the fear that data rules equal heavy regulation. The metrics it chose — failures, immobilizations, hard braking, miles travelled — are outputs a modern ADAS or AV stack already logs. Requiring their disclosure imposes minimal marginal cost while producing the public evidence base that proportionate, innovation-friendly regulation depends on. You cannot calibrate a rule to risk you refuse to measure.
What proportionate looks like for India
The pro-innovation path is not to copy California's enforcement teeth — geofencing directives and operator citations presume a deployed robotaxi fleet India does not have. It is to copy California's epistemics: standardize the data before the cars arrive.
- Finalize AIS-192 and tie it to a reporting duty. An EDR that records but never reports is a black box no one opens. Pair the standard with a public, anonymized aggregate-disclosure requirement modelled on vehicle-miles and failure reporting.
- Pick a small, output-focused metric set. Disengagements, system failures and miles-per-intervention are comparable across vendors and cheap to file. Resist bespoke Indian metrics that fragment from UN norms India already harmonizes with.
- Bind it to the DPDP Act, 2023. Vehicle telemetry is personal data. Aggregate safety reporting and individual privacy are reconcilable — but only if designed together, not bolted on after a breach.
India is investing heavily in the compute to run this future, targeting 100,000 AI accelerators by year-end. The missing piece is not silicon or even safety hardware — it is the boring, essential transparency layer that turns autonomous-vehicle deployment from an act of faith into an evidence-based policy. California spent its political capital building exactly that. India should borrow the blueprint, not the permit fees.