1. Headline & intro
The gig economy has found a new product to sell: not your time, not your car, but the world as seen through your smartphone camera. DoorDash’s new Tasks app turns its vast network of couriers into a roaming data-collection army for artificial intelligence and robots. This isn’t just another feature in a delivery app; it’s a glimpse of how AI companies plan to digitize reality itself—cheaply. In this piece, we’ll unpack what DoorDash is actually doing, why it matters for workers and AI, and what it signals for similar platforms in Europe and beyond.
2. The news in brief
According to TechCrunch, DoorDash has launched a standalone app called Tasks that pays its delivery couriers to complete small assignments designed to train and evaluate AI and robotic systems.
Couriers can earn extra money by recording videos of everyday activities—such as washing dishes while wearing a body camera—or by capturing themselves speaking in other languages. DoorDash says this data is used to help AI and robots better understand physical environments and human behaviour.
Bloomberg, cited in the reporting, notes that the footage and audio will be used both for DoorDash’s own AI models and for external partners in sectors like retail, insurance, hospitality and technology. In parallel, “digital tasks” are also being integrated into the existing Dasher app, such as taking photos of restaurant dishes, hotel entrances or manually closing doors on Waymo self‑driving cars.
The program is live only in select parts of the U.S., and notably excludes California, New York City, Seattle and Colorado—jurisdictions with tougher gig‑work rules. DoorDash plans to expand to more regions and task types over time.
3. Why this matters
DoorDash has quietly reframed what its asset really is. It’s not just a logistics network; it’s an on‑demand sensor network with eight million people who can be dispatched to capture almost any slice of the physical world.
Who benefits?
- DoorDash and AI buyers gain access to highly specific, real‑world data that would otherwise be slow and expensive to collect through traditional data‑labeling firms or in‑house teams. Need hundreds of videos of different sinks, kitchens or hotel lobbies? Push a notification, wait a few hours.
- AI and robotics companies get better training data for embodied AI, warehouse automation, domestic robots and vision‑language models. The dish‑washing task, for example, is textbook training material for home robotics.
- Couriers get a new earning stream that doesn’t depend on demand spikes or tips. For some, this will be a welcome way to monetize downtime.
Who loses—or takes on the risk?
The real cost is being shifted onto gig workers in subtle ways:
- Data rights and privacy. When couriers film their homes, workplaces or public spaces, they’re not just giving away their own biometric data; they may also capture bystanders or private interiors. How informed is consent here? How long is data stored? Can workers ever revoke it?
- Invisible labour. This is essentially data‑annotation work repackaged as “flexible earning.” Unless pay scales are generous, it risks repeating the underpaid “ghost work” dynamics seen on Amazon Mechanical Turk and similar platforms.
- Regulatory arbitrage. The absence of the program in California, NYC, Seattle and Colorado is telling. These are precisely the places that have pushed hardest on classifying gig workers as employees or imposing minimum pay standards. The implication: DoorDash wants to experiment where labour rules are softer.
In short, Tasks is a clever business move—but it deepens the power imbalance between platforms and workers, turning the latter into raw material for AI with limited control over how their contributions will be used.
4. The bigger picture
DoorDash is not operating in a vacuum. This move lands at the intersection of three accelerating trends.
1. The next wave of data collection for AI
Generative AI and robotics systems are starving for diverse, high‑quality, real‑world data—especially video and multimodal interactions. Synthetic data and web scraping only go so far; to train robots to operate in messy human environments, you need messy human environments.
DoorDash’s Tasks app is an industrialized version of something tech companies have done piecemeal for years: asking users to photograph storefronts, scan receipts, correct map data or label images. What’s new is the scale and precision: millions of couriers can be directed like a field research team.
2. Platform work as AI infrastructure
Late last year, Uber announced a similar plan to let drivers earn by performing micro‑tasks such as uploading photos to train AI models. Data‑labeling companies like Scale AI and Surge have already turned annotation into an outsourced industry, heavily reliant on workers in emerging markets.
DoorDash is fusing these worlds: the gig worker isn’t just a driver or courier; they’re simultaneously a data collector, annotator and test harness for AI systems. The more platforms copy this model, the more “ordinary” service work becomes inseparable from AI infrastructure.
3. From maps to the ‘digital twin’ of reality
Big Tech has long sought to build increasingly detailed digital models of the world: Google Street View, Apple’s Look Around, indoor mapping, LiDAR scanning for AR. DoorDash’s approach is more bottom‑up. Instead of expensive mapping cars, it mobilises workers carrying commodity smartphones and, in some cases, body cameras.
If successful, this becomes a template. Any company with a large distributed workforce—delivery, ride‑hailing, home services, even retail chains—will be tempted to monetize employees and contractors as roaming data‑harvesters.
5. The European / regional angle
From a European perspective, DoorDash’s Tasks may feel distant—it’s a U.S. rollout, and DoorDash’s consumer presence here is limited. But the company owns Wolt, a major player in the Nordics, Central Europe and parts of Southern Europe, and rivals like Glovo, Delivery Hero (Lieferando), Just Eat Takeaway and Bolt are watching closely.
Three European‑specific issues stand out:
GDPR and privacy. Video of dishwashing in your flat in Ljubljana, Berlin or Madrid can reveal faces, addresses, family members, even religious or health information. Under GDPR, that’s sensitive personal data. Any European rollout would require robust consent flows, clear legal bases for processing, strict purpose limitation and options to withdraw consent.
EU AI Act compliance. The AI Act places heavy emphasis on data governance, especially for high‑risk systems and foundation models. Companies that train models on crowd‑captured footage may need detailed documentation about data provenance, labelling procedures and bias mitigation. “We paid someone to film it on their phone” will not be enough.
Platform‑work regulation. The upcoming EU Platform Work Directive aims to curb bogus self‑employment and increase transparency in algorithmic management. If a courier is instructed, rated and potentially penalized based on their performance on AI tasks, regulators may view that as further evidence of de facto employment.
For European startups, there is also an opportunity: differentiate by offering ethically collected, well‑governed datasets, or by building tools that allow workers to retain more control and visibility over how their data trains AI.
6. Looking ahead
Expect three developments over the next 12–24 months.
Rapid copy‑pasting across platforms. If Tasks proves economically attractive—i.e., if AI buyers pay more for data than DoorDash pays workers—similar programs will spread to Uber Eats, Wolt, Glovo, Deliveroo, Bolt and beyond. The marginal cost of adding “task rails” to existing apps is low.
Regulatory and reputational pushback. Watch for:
- State‑level investigations in the U.S., especially in jurisdictions already wary of gig‑economy abuses.
- Early complaints in Europe once any pilot touches EU soil, centring on GDPR, worker exploitation and dark‑pattern consent.
- Campaigns by labour groups arguing that if couriers are now crucial to AI infrastructure, they deserve more than piece‑rate micro‑payments.
A new frontier of worker organizing. Data‑labeling has been largely invisible, concentrated in low‑income countries. DoorDash is bringing a version of that work into wealthy cities. That increases the odds of collective action—from attempts to set minimum pay per task, to calls for a “data dividend” or even residuals when models trained on workers’ contributions generate profits.
Key questions to watch:
- Will DoorDash publish pay ranges and average time per task, or will workers have to reverse‑engineer whether it’s worth it?
- Will AI buyers demand stronger proof that data is ethically sourced, pushing prices—and worker pay—up?
- Could regulators force a separation between “delivery work” and “data work,” each with different protections and rights?
7. The bottom line
DoorDash’s Tasks app is a strategically brilliant way to monetize an existing workforce as AI infrastructure—but it pushes gig workers deeper into a gray zone of precarious, opaque digital labour. If this model spreads, millions of people will be training AI systems without meaningful say in how their data is used or how value is shared. The real question for policymakers and users alike is simple: if workers are helping build the next generation of AI, how much ownership—financial and otherwise—should they have in what they create?
