Property tax growth is capped. Ticket quotas are a crime. Red light cameras are banned. Code enforcement is none of those things  which is exactly where the cameras went.

Second of two parts. Part One examined Dallas’s AI “blight score” program: 100 cameras on 50 sanitation trucks, more than 21,000 properties flagged in four months, and a pre-launch goal of 5,200 code enforcement notices a year.

Illustration by NexfinityNews. AI-generated composite created to depict the fiscal substitution described in this article. It is not a photograph of any real property, municipal building, or document, and does not depict any real person or address.

There is a number in the Dallas record that should have ended the conversation before it started.

Asked what its AI camera program was meant to accomplish, the city said that before implementation, its only measurable goal was the issuance of 5,200 courtesy notices per year following full rollout.

Now consider a second fact. In Texas, a city may not establish or maintain, formally or informally, a plan to “evaluate, promote, compensate, or discipline” a peace officer based on issuing a predetermined number of traffic citations. It may not even suggest that an officer is expected to hit such a number. Violation is misconduct and grounds for removal from office. A former Texas city administrator was convicted of abuse of official capacity under that statute  in a municipality where fifty-four percent of the 2017-18 budget came from traffic fines.

Texas decided, decades ago, that a government setting numeric enforcement targets is corrupt in principle. Then it wrote that principle into law for cars only.

That gap is not an oversight. It is the whole architecture of modern municipal finance, and Dallas’s trash trucks are simply the newest instrument pointed through it.

The squeeze is real, and it is deliberate

Texas capped local property tax growth in 2019. Senate Bill 2 lowered the multiplier in the voter-approval tax rate calculation from eight percent to 3.5 percent for nearly every Texas city, meaning revenue growth above that line requires an election.

The legislature tried to tighten it further last year and failed. SB 10, filed in the 2025 second special session, would have dropped the cap to 2.5 percent for jurisdictions over 75,000 population. House members amended it down to one percent for all local governments with a public safety carve-out; the Senate stripped those changes; the House then rejected the conference report and the bill died. The 3.5 percent cap stands, and every Texas city budget office now assumes the ratchet turns again next session.

Dallas is operating inside that vise. The city’s FY 2026-27 recommended budget runs roughly $5.66 billion against a previously projected $50.9 million General Fund shortfall. It eliminates 296 budgeted positions. It cuts the property tax rate for the eleventh consecutive year. Sales tax, roughly a quarter of the General Fund, has been coming in under projection, and a 2024 charter amendment claims half of all new revenue for police and fire.

Set those facts beside one another and the incentive is not subtle. Property tax growth is capped and politically radioactive. Sales tax is soft. Half of anything new is spoken for. What remains is the category of money a city can increase without asking anyone: fines, fees, penalties, abatement billings, and liens.

The person who built the cap predicted this outcome in public. Writing in defense of SB 2 in July 2019, Lieutenant Governor Dan Patrick noted that some local officials were already looking for ways to maneuver around the law, including proposals to eliminate the homestead exemption and “threats to slap excessive fees on residents.” He promised the legislature would close such loopholes.

Seven years later, the loophole is a fleet of garbage trucks with cameras on them.

This is a studied phenomenon, not a theory

The pattern has a name in the academic and policy literature: taxation by citation.

The Institute for Justice examined three Georgia cities and found that from 2012 to 2016, fines and fees produced annual average revenue shares of 14 percent for Riverdale, 17 percent for Morrow, and 25 percent for Clarkston  against three percent in comparably sized Georgia cities. In each, fines and fees were the second largest revenue source after property taxes. Most citations had little to do with public safety.

A 50-state analysis published in June 2026 found 275 jurisdictions across 25 states reporting fines exceeding ten cents for every dollar of general revenue in their 2023 fiscal years. New York leads per capita at $75.81 per resident against a national weighted average of $24.77.

The extremes are instructive because of what they provoked. Brookside, Alabama  population roughly 1,250  grew its fines and forfeitures revenue by more than 640 percent between 2018 and 2020, until fines accounted for half the town’s budget. Alabama responded in 2022 with a ten percent cap, the strictest in the nation, that applies only to traffic citations. Missouri’s Macks Creek Law caps municipal ordinance and minor traffic fines at 20 percent of general operating revenue and has been revised repeatedly across three decades in response to persistent evasion, including after the Ferguson investigation exposed its limits.

Note the shape of every one of those reforms. A scandal happens. A cap follows. The cap covers the instrument that caused the scandal. Enforcement migrates to the instrument the cap does not cover.

Texas has regulated this three times  always for cars

This is the part that deserves to be read slowly, because Texas has been unusually aggressive about revenue-driven enforcement and unusually narrow about where the rules apply.

One: a revenue cap that exempts almost every city. Transportation Code § 542.402  the “speed trap” law  lets a municipality retain traffic fine revenue up to 30 percent of the prior year’s total revenue from all sources other than federal funds and bond proceeds. Above that, everything over a dollar per citation goes to the state comptroller. Jurisdictions collecting at least 20 percent must file an annual report. But the cap applies only to municipalities under 5,000 population. Dallas, at roughly 1.3 million, is not covered by any fines revenue cap at all. And because the statute does not exclude enterprise revenue such as water and sewer funds from the denominator, even covered towns can pad the base to stay under the line.

Two: an anti-quota law that stops at the curb. § 720.002 applies to every political subdivision regardless of population. No city may evaluate, promote, compensate or discipline a peace officer by the number of traffic citations issued, or a judge by the money collected from traffic convictions, and no city may require or suggest such an expectation. It is enforced seriously: the Sixth District Court of Appeals in Texarkana upheld an abuse-of-official-capacity conviction against a former city administrator, holding that the statute reaches any official who suggests a quota to a peace officer regardless of whether that official has direct authority over them.

But the statute defines “traffic offense” by reference to Chapter 521 and Subtitle C of the Transportation Code. Driver’s licenses and rules of the road. A code compliance officer is not a peace officer. A courtesy notice for peeling paint is not a traffic citation. A stated annual target of 5,200 notices sits entirely outside a law written to forbid exactly that logic.

Three: an outright ban on the cameras  the traffic ones. HB 1631 prohibited red light cameras statewide effective June 2, 2019, and barred the use of photographic enforcement evidence to bring a charge or citation. Legislative analysis at the time put statewide red light camera revenue at roughly $39.6 million a year and noted plainly that ending the cameras would remove the incentive for cities to issue tickets to raise revenue. The bill also prohibited county tax assessor-collectors from refusing vehicle registration over unpaid red light camera fines.

Now hold those three next to the Dallas program.

Automated camera enforcement of traffic law: banned. Automated camera enforcement of property condition: unregulated. Numeric enforcement targets for peace officers: criminal misconduct. Numeric enforcement targets for code officers: a line in a budget briefing. Registration holds for unpaid camera-generated traffic fines: prohibited by statute. Registration holds for unpaid code fines: alive and operating, through the same scofflaw program the legislature closed to red light cameras seven years ago.

Every principle Texas has articulated about revenue-driven automated enforcement applies to what Dallas is doing. None of the statutes do.

So who actually writes the rules being enforced?

If a camera can now find violations at industrial scale, the next question is who decides what counts as a violation. The answer runs through three layers, and they are not equally accountable.

Layer one: a private nonprofit. Most American building codes descend from the model I-Codes published by the International Code Council, a private organization whose documents carry no legal force until a government adopts them. The ICC runs a three-year development cycle. Some code development committee members are nominated by industry partners and approved by the ICC Board. Final votes on proposed changes are cast by the ICC’s governmental members  public code officials  through committee action hearings, public comment hearings, and a two-week online governmental consensus vote. It is a real consensus process with a real industry channel running alongside it.

Layer two: the state. Texas does not leave adoption to chance. Local Government Code § 214.212 adopts the International Residential Code as the municipal residential building code statewide, and § 214.216 adopts the International Building Code for commercial structures. Cities may adopt local amendments that add, modify, or remove requirements  but only after a public hearing and only by ordinance.

Layer three: a Wednesday morning. And here is the finding that matters.

None of that governs blight enforcement.

High grass. Litter. Peeling paint. Junk vehicles. Outside storage. Debris. Deteriorating structures. These are not building code violations. They live in local property maintenance and nuisance ordinances  in Dallas, Chapter 27 “Minimum Property Standards” and Chapter 18. The ICC publishes a model International Property Maintenance Code that many communities fold into their housing codes, but nothing requires its adoption and nothing constrains its content. No three-year cycle. No state floor. No governmental consensus vote. No public comment hearings in Louisville.

A city council majority, on any given agenda.

The codes the cameras enforce are precisely the codes with the least outside process in American local government. That is not a coincidence  nuisance law has always been local because nuisance is local. It becomes a problem only when enforcement capacity goes from fifty properties an inspector can walk to a fleet that photographs every parcel in the city on a thirty-day cycle.

Will the AI start writing codes?

Almost certainly not, and the strongest version of this argument does not require it.

Credit where due: City Detect’s published Responsible AI Strategy commits to taking “No revenue from detections,” stating the company never collects revenue based on the number of citations issued. That is a direct answer to the crudest version of the concern, and it is more than most surveillance vendors put in writing.

The realistic mechanism is a ratchet, not a robot legislator, and it has three teeth.

Detection classes are chosen, not discovered. Someone decides what the model looks for. That decision is upstream of every notice and it is made jointly between a vendor and a client department, not by ordinance. Adding a detection class requires no vote at all.

What is cheap to see becomes what gets enforced. A camera at street level can flag chimney discoloration, a leaning fence, an inoperable vehicle. It cannot flag a failing furnace, a bad panel, mold behind drywall, or a landlord refusing heat. Over time, an enforcement regime optimized around machine-visible conditions drifts toward the cosmetic and away from the dangerous. The exterior of a poor person’s house becomes the enforcement priority; the interior of a slumlord’s building does not.

The data justifies the next ordinance. Dallas’s own briefing said the imagery would also serve planning, neighborhood outreach, and grant reporting. The vendor’s Prescott Valley case study is titled around grant compliance and reports 712 undeveloped residential lots and 4,158 blight indicators. Once a city holds a parcel-level condition index, it holds the evidentiary basis to argue for new overlay districts, new standards, new grant applications, and new enforcement. The camera does not write the code. It manufactures the finding that makes the next code look necessary.

None of that is sinister. All of it is a one-way ratchet with no countervailing pressure, in the one area of municipal law with no external process.

What reform would actually look like

The useful proposals are narrow and mostly involve extending rules Texas already wrote.

1.     Extend § 720.002 beyond traffic. If numeric enforcement targets are corrupt when applied to peace officers, they are corrupt when applied to code officers. A one-line amendment covering municipal ordinance enforcement would have made Dallas’s 5,200-notice goal unlawful on its face.

2.     Extend a fines revenue cap to all municipalities and all fines. The 30 percent limit in § 542.402 exempts every city over 5,000 people and covers only traffic. Both limitations are indefensible on the logic that produced the statute. Close the enterprise-fund loophole in the denominator while you are there.

3.     Require disclosure of detection classes. Any automated enforcement system should have to publish, by ordinance, what it is configured to detect, with changes requiring the same public hearing that a local building code amendment requires under § 214.212.

4.     Mandate a retention and access schedule. How long images are kept, who may query them, under what predicate, with what audit log, and whether outside agencies may request access. Dallas has published none of this, and the vendor’s six public commitments do not include a retention period.

5.     Ban notice counts as a program metric. Measure compliance achieved, hazards abated, and repairs completed. If a program cannot show those numbers, it is not a code enforcement program.

6.     Attach a repair pathway to every camera-generated notice. A notice that finds a violation but offers no route to fixing it is a collection instrument aimed at people whose only offense is being unable to afford maintenance.

The question underneath all of it

Cities need revenue. Dallas is not inventing a shortfall, and code enforcement is not inherently illegitimate  overgrown lots and collapsing structures are real problems that fall hardest on the neighbors who live beside them.

But there is a difference between a city that enforces its codes and a city that has discovered enforcement is the only line in its budget that voters cannot cap. The first responds to conditions. The second responds to the gap between what it wants to spend and what it is allowed to collect.

The distinguishing test is not the technology. It is what the program measures. A city that counts abated hazards is doing code enforcement. A city that counts notices issued has told you what the notices are for.

Dallas counted notices. It said so before the first camera was mounted.

Frequently asked questions

Are ticket quotas illegal in Texas?

Yes, for traffic. Transportation Code § 720.002 bars any political subdivision from evaluating, promoting, compensating, or disciplining a peace officer by the number of traffic citations issued, or a judge by fines collected from traffic convictions, and from requiring or suggesting such expectations. Violation is misconduct and grounds for removal.

Does that law cover code enforcement?

No. The statute defines traffic offenses by reference to Chapter 521 and Subtitle C of the Transportation Code. Code compliance officers are not peace officers and property maintenance notices are not traffic citations.

Is there a limit on how much of a Texas city’s budget can come from fines?

Only for very small cities. Transportation Code § 542.402 caps retained traffic fine revenue at 30 percent of prior-year revenue, but applies only to municipalities under 5,000 population, and only to traffic fines. Dallas is subject to no fines revenue cap.

Why did Texas ban red light cameras but allow code enforcement cameras?

The 2019 ban addressed due process and revenue incentives in automated traffic enforcement specifically. Nothing in it reaches automated detection of property conditions, which was not a meaningful technology at the time.

Who writes property maintenance codes?

Local governments. Building codes descend from ICC model codes adopted by Texas statute with local amendments requiring a public hearing. Property maintenance and nuisance provisions  the ones cameras detect  are adopted by local ordinance with no state-mandated model and no external consensus process.

Can unpaid code fines still block my vehicle registration?

Yes. The scofflaw framework that Texas closed to unpaid red light camera fines remains available for other past-due municipal fines and fees.

Related coverage

·       Scored by a Garbage Truck: Inside Dallas’s AI “Blight” Surveillance Program (Part One)  

·       When Cities Cancel Flock: The Municipal Backlash Against ALPR Surveillance

·       The Driver Data Economy, Part One: How Your DMV Record Became a Product

·       Axon’s Second Act: The Company Buying Up the Surveillance Market It Once Funded