Uttar Pradesh Notified Aggregator Rules. Platforms Now Need a Cover Map That Works in Three States at Once

On 21 August 2026, ThePrint reported that Uttar Pradesh has notified rules for app-based transport aggregators that cap dynamic fares at 50 per cent above the base fare and bring delivery services within the regulatory perimeter. The Times of India carried the same development under a blunter headline: app-based cabs can no longer charge an arbitrary fare in UP. The Free Press Journal reported the framework on 22 August 2026 as the state's aggregator policy for 2026.

The licensing piece landed earlier. Hindustan Times reported on 29 June 2026 that Uttar Pradesh has made operating licences compulsory for app-based cab and delivery operators. The News Mill, reporting on 21 August 2026, framed the package as rules for app-based rides and deliveries together, which is the part that matters most for insurance. A delivery platform that previously sat outside transport regulation in the state now sits inside it.

Two features of the UP framework drive the cover question rather than the pricing question. First, the obligation runs through a licence held by the platform, which fixes who the regulator will hold answerable after an incident. Second, delivery is in scope alongside ride-hailing, so a quick-commerce operator running rider-owned two-wheelers in Lucknow, Kanpur and Noida is now a licensed entity in India's largest state rather than an unregulated marketplace.

UP does not arrive into a vacuum. Two other state regimes are live or nearly live, and they attach obligations at different points.

1. Karnataka applies a levy on transactions. MediaNama reported on 10 August 2026 that the state applies a 1 per cent levy per transaction, capped at 50 paise for food and grocery deliveries. That levy is under an Article 254 challenge on the ground that it conflicts with the central Code on Social Security, 2020, and the welfare fee has been running into a court registry rather than the welfare board while the challenge is heard.
2. Maharashtra has a plan but nothing notified. The Indian Express reported on 8 August 2026 that Swiggy, Zomato and Zepto face an EV mandate and a rider welfare levy under the Maharashtra plan. Until it is notified, the exposure is a planning exposure.
3. Uttar Pradesh attaches the obligation to a licence and to fare conduct, with delivery inside the scope.

The practical consequence is that the same rider, on the same vehicle, running the same trip, can sit under a transaction-based levy in Bengaluru, a licence condition in Lucknow, and a draft welfare charge in Pune. None of these regimes cross-recognises another. A platform that treats state compliance as a single national workstream will discover the divergence at the worst moment, which is when a claim, an inspection or a licence renewal forces it to evidence what applies where.

The first axis to compare across states is who the regulated person is, because that determines the named insured on every policy the platform buys.

Under the UP framework the licence sits with the aggregator, per the Hindustan Times report of 29 June 2026 that licences are compulsory for app-based cab and delivery operators. That makes the platform entity the party a regulator can act against, and it makes the platform entity the party a claimant will name first. The platform, not a fleet partner and not a rider, must therefore be the first named insured on the liability programme, with fleet partners and vehicle owners added by endorsement where the contract requires it.

This is not a cosmetic drafting point. Where the platform operates through a separate state subsidiary, a delivery entity distinct from the mobility entity, or a franchise structure for dark stores, the licensed entity and the insured entity can diverge. A certificate of insurance issued in the name of the group holding company does not evidence cover for a licensed operating subsidiary in UP.

The checks worth running before the next renewal:

• List every legal entity that holds or will apply for an aggregator licence in any state, including delivery-only entities now caught by the UP rules.
• Confirm each of those entities appears on the liability schedule, either as a named insured or through a clearly worded subsidiaries clause.
• Confirm the policy wording definition of insured operations covers delivery as well as passenger transport, because a mobility-drafted wording may describe carriage of passengers only.
• Keep entity-level certificates ready in a form a state transport authority will accept at licence renewal.

The second axis is whether the state fixes the duty on the platform or on the person who owns the vehicle. It changes what insurance actually responds.

Motor third-party cover is statutory under the Motor Vehicles Act and attaches to the vehicle. In an aggregator fleet built on rider-owned two-wheelers and driver-owned cars, that cover sits on assets the platform does not own and cannot unilaterally renew. When a state makes the operating licence compulsory for the platform, the platform inherits a supervision duty over an insurance obligation that legally belongs to someone else.

The workable answer is to treat motor third-party verification as a licence condition rather than an onboarding formality:

• Capture policy number, insurer and expiry at onboarding for every vehicle attached to the platform, and store the document, not just the number.
• Run an automated expiry check that blocks allocation when cover has lapsed, rather than a monthly report that someone reviews after the fact.
• Re-verify at intervals, because a rider can cancel or fail to renew after onboarding.
• Keep an audit trail per vehicle, because the licence-holder is who a transport authority will ask to produce it.

What motor third-party does not do is respond to platform-level allegations. If a claimant argues that the algorithm pushed unsafe delivery times, that the onboarding checks were inadequate, or that the platform failed to supervise a driver it had already received complaints about, the claim is against the platform's own conduct. That belongs on a public liability layer, discussed below, and the exposure is the same one covered in our earlier note on aggregator liability for gig workers.

The third axis is the hardest to draft around, and it is the one that produces disputed claims. Riders multi-app. A two-wheeler rider can be logged into two delivery platforms and a ride-hailing app in the same evening, and an accident happens at a single point in time on a single trip.

Cover written on an on-duty basis asks whether the rider was engaged on that platform's trip at the moment of loss. That definition is precise and cheap, and it fails in exactly the multi-app case, because the rider may have been between orders, riding to a pickup for another app, or logged in without an active task. Each platform's insurer can then point at the other, and the rider's family waits while two adjusters argue about a timestamp.

Cover written on a 24-hour basis removes the argument entirely. Every registered rider is covered around the clock, the trigger is registration rather than task state, and the contribution principle handles overlap between two platforms' policies without leaving the rider uncovered. It costs more per life, and it is the version that survives contact with a real claim.

If cost forces an on-duty structure, define on-duty to include the period from login to logout, including the interval between tasks and the ride to and from the first and last task of a session. Then confirm in writing that the insurer accepts platform login logs as proof of duty status, because that log is the only evidence that will exist after a fatal accident.

The multi-app problem also affects sums insured. A rider with cover from three platforms does not have three times the benefit if each policy carries a contribution clause. Tell riders what their actual expected recovery looks like rather than letting them assume the covers stack.

The fourth axis is how each state's welfare charge sits against the central Code on Social Security, 2020, which itself contemplates an aggregator contribution into a national fund for gig and platform workers.

Karnataka's design, per MediaNama's 10 August 2026 report, is a 1 per cent levy per transaction capped at 50 paise on food and grocery deliveries, and it faces an Article 254 challenge on the argument that a state levy conflicts with the central Code. Article 254 governs repugnancy between central and state law on concurrent subjects, so the challenge goes to whether the state can occupy this ground at all while the Code exists. Maharashtra's plan, reported by The Indian Express on 8 August 2026, pairs a rider welfare levy with an EV mandate and has not been notified.

Three consequences for a platform finance and risk team:

1. Do not net one against the other in planning. Until the Article 254 question is decided, assume both a state charge and a central contribution can apply, and budget accordingly. If the challenge succeeds, the provision releases. If it fails, the provision was correct.
2. A welfare charge is not insurance. A levy funds a state board, on a state timeline, subject to registration of the worker. It does not pay a claim to a rider's family in the weeks after an accident. The insured layer does that. We covered the accounting posture for a levy sitting in escrow in the note on the Karnataka welfare fee litigation.
3. Tag transactions by state at source. Every state charge is computed on that state's transaction base. A platform that cannot split transactions cleanly by state cannot compute any of these charges defensibly, and it will face the same problem each time another state legislates.

The mistake to avoid is buying three programmes. State frameworks diverge on triggers, but the underlying exposures are the same everywhere: a rider gets hurt, a third party gets hurt, a vehicle is uninsured, or the platform's own conduct is challenged. Build one national programme and evidence it state by state.

Four layers do the work:

1. National group personal accident, written on a scheme basis. Cover attaches to registered riders as a class rather than a named list, so daily churn does not create gaps. Capital sums for death and permanent total disability, with permanent partial disability scaled to a table, and a declaration mechanism that reconciles headcount periodically instead of requiring per-rider endorsements.
2. Accident medical expenses as a distinct benefit. Most rider accidents are not fatalities. They are fractures, ligament injuries and hospital admissions of a few days. A personal accident capital sum pays nothing for these unless a medical extension sits alongside it, and this is the benefit riders actually claim on.
3. Public liability for platform-level claims. This responds to third-party bodily injury and property damage arising from platform operations, including allegations aimed at the platform's own systems and supervision rather than at a single rider's negligence. It sits above motor third-party cover.
4. Motor third-party verification as a controlled process, treated as a licence condition on rider-owned vehicles, with evidence retained per vehicle.

Run all four nationally, with limits set on the whole exposure. A single group personal accident scheme priced on total rider count will almost always beat three state programmes priced on fragments, and it removes the cross-border gap that appears the moment a rider relocates from Noida to Bengaluru.

The structure that makes one programme survive three regulators is a state annexure: a short document per state, sitting on top of the national policy, that maps each state requirement to the layer that satisfies it and to the evidence a compliance team can produce on request.

Each annexure should record, for one state:

• The licensed entity or entities operating there, and the named-insured position for each.
• The state's specific obligations as notified, with the source and date, so a reviewer can see what the annexure is answering.
• Which national layer meets each obligation, with the policy number, period and sum insured.
• The evidence artefact for each: the certificate, the scheme declaration, the vehicle verification report.
• Any state charge payable, its computation base, and where it is paid or deposited.

This is deliberately boring, and that is the point. When a UP transport authority asks a licensed delivery operator to demonstrate what cover its riders carry, the answer should be a one-page annexure and a certificate, not a scramble through a broker's inbox. The same annexure structure absorbs Maharashtra when its draft is notified, without renegotiating the underlying programme.

The timing argument is straightforward. UP is notified now, Karnataka is live and litigated, Maharashtra is coming. A platform that builds the entity list, the national four-layer programme and the annexure structure during this renewal cycle is ready for the fourth state. One that waits will be doing entity mapping under a licence-renewal deadline.