Oakland’s city government has taken the first formal step toward contracting Aerbits, a San Francisco-based startup, to operate AI-equipped drones across the city’s most dump-prone corridors in a six-month, $150,000 pilot program. The Oakland City Council approved the measure on a first reading in April 2026, with a second reading required before the contract becomes final. If enacted, the program will deploy 72 drone missions covering approximately 1,440 road miles, targeting between five and ten square miles identified as the city’s worst illegal dumping hotspots.
A City Spending $24 Million a Year on a Problem It Cannot See
Oakland Public Works collects an estimated 6,000 pounds of illegally dumped material every day, and the city spends around $24 million annually managing the problem. The $150,000 Aerbits contract would represent roughly 0.6% of that annual expenditure. Kristen Hathaway, Assistant Director of Oakland Public Works, told councilmembers the technology was expected to make crew dispatch substantially more efficient, allowing the city to route the right equipment to the right locations rather than relying on a patchwork of resident complaints to guide operations.
The core limitation of the current complaint-based model is structural. Neighborhoods that generate the fewest 311 reports are frequently those with the worst dumping concentrations. District 2 Councilmember Charlene Wang described the disparity bluntly: some of the most heavily affected parts of her district, including the San Antonio and Little Saigon areas, produce only around five reports in a given reporting window, while wealthier parts of the same district generate roughly 400 reports for comparatively minor incidents.
How the Detection System Works
The Aerbits platform operates DJI Mavic 3 Enterprise drones flying at altitudes of 120 to 150 feet, where the aircraft’s cameras photograph streets and sidewalks continuously. The M3E carries a 20-megapixel Four Thirds CMOS sensor with a mechanical shutter, which eliminates motion blur during grid-pattern mapping passes, and can trigger an image every 0.7 seconds. With an optional RTK module, the aircraft achieves centimeter-level positional accuracy. Maximum flight time reaches 45 minutes per battery cycle.
Onboard computer vision software classifies waste by type, including mattresses, tires, furniture, and construction debris, then measures pile dimensions using centimeter-grade GPS data. Once a dumpsite is confirmed, the system automatically generates a 311 service request with timestamped aerial imagery and precise coordinates attached. No facial recognition is applied, and no license plate data is collected. Under Oakland’s proposed data retention rules, original unredacted images would be held for one week, while redacted versions would be retained for six months.
At operational pace, a single drone covers approximately one square mile every 30 minutes. That rate is substantially faster than ground-based code enforcement, which typically covers a few city blocks in the same timeframe.
Evidence From San Francisco’s Bayview Neighborhood
Aerbits is not presenting this as an untested concept. The company’s founder, Brian Johnson, a software engineer and Bayview-Hunters Point resident, ran a 13-month self-funded pilot in that San Francisco neighborhood between March 2022 and April 2023. Across 125 missions and nearly 118,000 aerial photographs, the program identified 4,441 discrete dumpsites and filed 4,376 automated service requests to San Francisco 311. At peak, active dumpsite counts fell by 96 percent relative to baseline.
The most analytically significant portion of the Bayview data came from a structured 52-day observation period in the spring of 2022 that followed an A-B-A withdrawal design. The baseline count on day one stood at 118 active dumpsites. After 26 days of daily flights, that figure fell to 85. Monitoring was then suspended for 14 days, during which active dumpsites rebounded to 91. When flights resumed over a final 24-day period, the count dropped to 5.
The rebound during the suspension interval is the critical data point. Random variation, seasonal factors, or concurrent city cleanup efforts could explain an initial reduction. They cannot easily explain a return toward baseline when the only variable that changed was whether the drone was flying, followed by a second sharp decline when it resumed.
One additional finding from Bayview carries significant implications for how cities understand their own dumping problem: between 30 and 50 percent of the dumpsites the drone identified on public roads and sidewalks had no corresponding 311 report at all. The gap reflects the multiple barriers to complaint-based reporting, including sites that are not visible from main thoroughfares, language access issues, and resident skepticism that reports lead to action. European cities deploying AI monitoring to address similar enforcement gaps have pursued comparable logic. Kurrant previously covered a programme in France’s Dordogne region where AI cameras were installed at communal waste points after traditional reporting mechanisms failed to capture the scale of fly-tipping activity.
Privacy Review and Programme Boundaries
Oakland’s Privacy Advisory Commission reviewed the Aerbits proposal and approved it in March 2026, before the matter advanced to the Public Works and Transportation Committee. Councilmember Zac Unger has been explicit that the programme’s scope is logistical rather than investigative: drones identify where garbage accumulates so crews can be dispatched with the correct equipment, not to identify or track individuals.
The programme sits alongside but is legally and operationally separate from a parallel city discussion about expanding licence plate reader infrastructure to identify individuals caught dumping. Oakland currently operates 36 licence plate cameras citywide. That initiative involves substantially different privacy implications and distinct oversight requirements. Oakland Privacy, a community advocacy group, has noted it will monitor the drone pilot closely, citing concerns about function creep and the potential for false positives.
Operational Sustainability and What the Data Actually Requires
The Bayview results carry a constraint that is directly relevant to how Oakland should plan this programme: the dumpsite reductions persisted only as long as flights continued. When monitoring paused for two weeks, dumpsites recovered toward pre-programme levels. That trajectory implies the deterrent and logistical benefits of aerial detection are continuous rather than cumulative. Any decision to treat a six-month pilot as a one-off experiment rather than a foundation for recurring operations would likely yield a temporary result rather than a sustained one.
Oakland has separately doubled its illegal dumping fines, with penalties now reaching $5,000 per incident and additional daily fines of up to $1,000 for commercial and hazardous waste offences. The fines are now tied to vehicle licence plates for the first time, creating an enforcement mechanism that complements detection. Whether that combination of higher penalties and aerial monitoring produces durable change in dumping behaviour over a longer horizon will be the central question when the pilot concludes.
