Dallas Garbage Trucks Use AI Cameras to Assign Homes 'Blight Scores' for Code Violations

AI Cameras Scan Dallas Properties from Garbage Trucks

Dallas began testing AI-powered cameras on garbage trucks in April, mounting 100 cameras in pairs across 50 brush and bulky-waste collection vehicles1

1

. Each truck features cameras pointing from both sides, capturing 1920 × 1080 resolution images at street level using fixed-focal-length lenses without optical or digital zoom. The system, operated by City Detect through a three-year, $2.5 million contract, is designed to provide a visual scan of Dallas roughly once every 30 days1

1

. Machine vision technology enables these AI cameras to automatically identify code violations including overgrown grass, litter, debris, overgrown weeds, and structural issues across neighborhoods1

1

.

Properties Assigned Blight Scores Based on Violations

The AI-powered cameras on garbage trucks have flagged at least one possible code violation on more than 21,000 properties since the program's expansion over the summer1

1

. Each property receives a blight score ranging from 1 to 4, with 1 representing the least severe violations and 4 indicating the most serious infractions. One homeowner received a score of 2 for a chimney paint violation, which they considered excessively harsh1

1

. The city has already issued 1,800 courtesy notices to property owners, with a goal of sending approximately 5,200 per year1

1

. These courtesy notices serve as warnings that could escalate to fines if violations remain unaddressed, effectively using AI to automate code enforcement across the city.

Privacy Concerns Mount Over Unauthorized Photography

Residents have raised significant privacy concerns about having their homes photographed without consent, questioning who defines what constitutes blight on their property1

1

. City Detect CEO Gavin Baum-Blake stated that faces and license plates are blurred in captured images, and the company maintains that images are captured in public spaces where there is no expectation of privacy, similar to Google Street View2

2

. Dallas officials emphasize that the cameras operate at street level and cannot see above fences1

1

. Images and data are stored in cloud storage located in the United States, maintained by City Detect for the contract's duration1

1

. The company insists it doesn't earn money on a per-citation basis and that images don't feed into law enforcement databases2

2

.

Lower-Income Neighborhoods Disproportionately Targeted

Zip codes with the highest number of civic code violations flagged by the AI cameras were concentrated in Dallas's lower-income southern quadrant2

2

. This pattern has raised concerns about the program disproportionately fining families already struggling financially in economically challenged areas. City councilmember Chad West characterized the initiative as potentially becoming "a tax on the poor," highlighting worries about regressive impacts on vulnerable communities2

2

. The concentration of violations in specific neighborhoods suggests the system may amplify existing inequalities rather than address underlying community needs. Watch for data showing whether enforcement patterns shift as municipal governments across the country adopt similar technologies.

AI Hallucinations Raise Accuracy Questions

The potential for AI hallucinations presents serious concerns about false detections and wrongful citations. A documented case from last year involved a 16-year-old high school student who was swarmed by police and handcuffed after an AI camera system mistook his bag of Doritos for a gun1

1

. Dallas claims its AI models were trained on examples covering different seasons, building styles, neighborhoods, lighting conditions, weather, and geographic areas to reduce inconsistent or biased detections1

1

. City Detect configures the AI according to Dallas's requirements, but questions remain about who validates the accuracy of automated assessments before courtesy notices are issued.

City Councilmember Proposes Budget Amendment to End Program

City councilmember Chad West proposed an amendment to strip funding for the camera contract, though he agreed to table his request after the city manager committed to a hearing on the issue in December1

1

. West acknowledged that the cameras could serve legitimate purposes, such as detecting illegal dumping sites, but emphasized the need for further investigation into their impact on families and who can access the images1

1

. The December hearing will likely determine whether Dallas continues its partnership with City Detect or joins the growing number of municipalities reconsidering AI-powered surveillance programs.

Broader Trend of AI Surveillance Faces Backlash

City Detect currently partners with at least sixteen other cities across the United States beyond Dallas2

2

. The company markets primarily to municipal governments with limited personnel and budgets, promising that AI can handle busywork so staff can focus on pressing tasks2

2

. However, Americans are increasingly wary of AI-powered surveillance and tech companies' capability to covertly gather personal data. Flock Safety, another surveillance company, has faced particularly strong opposition: Republican Senator Josh Hawley launched an investigation citing documented cases of police officers using automated license plate readers to track people not suspected of crimes, including ex-wives and ex-girlfriends2

2

. Over 200 counties and cities across the U.S. cancelled their contracts with Flock Safety since August 2021, with 166 cancellations occurring in a single month alone, according to data from Secure Justice2

2

. This pattern suggests Dallas's program may face similar resistance as residents and officials weigh efficiency gains against civil liberties concerns.