California has produced the kind of evidence road safety researchers have been waiting years for on automated speed enforcement.
According to an evaluation by the San Francisco Municipal Transportation Agency, extreme speeding fell by 72 percent at locations where speed cameras were installed.
The number matters. How it was measured matters more.
How the measurement was done
Before any cameras went up, SFMTA used pneumatic tubes to record vehicle speeds at all 33 planned locations, establishing a baseline. After installation, speeds were measured again at 15 of those locations in September 2025.
This is not a survey or a perception study. It is a before-and-after measurement at the same sites using the same method.
| Measure | Result |
|---|---|
| Reduction in extreme speeding (more than 10 mph over the limit) | 72% |
| Fall in average speed | 4 mph |
| Fewer speeding vehicles per day | Approximately 20,000 |
But the most striking finding is not in the table.
Before installation, average speeds exceeded the posted limit at 12 of the 15 sites. Afterwards, average speed was below the posted limit at all 15.
That result speaks directly to the AAA Foundation research published this month, which found that drivers take their cues from surrounding traffic rather than the sign, and do not expect enforcement until roughly 10 to 12 mph over the limit. The San Francisco data shows what happens when that expectation changes.
SFMTA launched the programme on 20 March 2025, began issuing warnings in June 2025 and citations in August 2025.
Oakland: the second city, a year behind
California’s pilot programme, authorised under Assembly Bill 645, put San Francisco first and Oakland second. Los Angeles, San Jose, Glendale and Long Beach are also participating.
Oakland switched on 35 cameras at 18 locations on 14 January 2026, running a warning period before citations began on 15 March. The cameras photograph the rear plate of vehicles exceeding the limit by 11 mph or more. Fines range from $50 to $500.
Oakland has not yet published before-and-after speed data; OakDOT has said a fuller analysis is due later this year. What it has released shows the scale of the underlying problem:
- 140,445 speeding violations in the first five weeks, an average of 3,601 per day
- Roughly 82,000 citations in the first full month of enforcement
- Average speeds of 55.3 mph on Hegenberger Road, more than 15 mph above the limit
The sites were not chosen at random. They were selected from crash history and speed studies along Oakland’s High Injury Network, where some locations recorded more than 10,000 vehicles a day travelling at least 11 mph over the limit.
Two people are killed or seriously injured in traffic in Oakland every week.
The most useful figure in the dataset
One finding from Oakland’s warning period says more than any citation total.
Two thirds of warnings went to drivers who had already received two or more.
This is what the speed camera debate should be organised around. The problem is not that everyone occasionally drifts over the limit. It is that a comparatively small group of drivers exceeds it systematically and repeatedly. High citation volumes look like a revenue operation until you see that the volume is concentrated rather than distributed.
It also points to where policy should go next. For drivers a single fine does not deter, the answer is graduated sanctions and repeat-offender programmes, not more cameras.
The objection: “a money grab”
As cameras spread across the US, so does driver resistance. They are now legal in more than two dozen states, with Florida, Hawaii, Michigan and Minnesota joining California in recent years.
An NPR report on 16 August documented the backlash in Florida, where drivers ticketed by school-zone cameras describe the programme as revenue collection. The specific complaint is that citations are issued during the middle of the school day, when children are inside the building rather than arriving or leaving.
A law firm challenging camera tickets there argues that a genuinely safety-driven programme would operate only when children are present.
This is a technically answerable objection, and the answer lies in programme design: do the camera’s operating hours match the hours during which the school-zone speed limit is legally in force? If they do not, the complaint has merit. If they do, the driver has violated a different limit that was in effect. It varies by programme, and it is worth checking rather than assuming.
What the safety side says
Raul Avelar of the Insurance Institute for Highway Safety frames the purpose narrowly: cameras reduce speeds, so crashes that do happen occur at lower speeds, and fewer people die.
He points to multiple US studies showing cameras cut speeding by 70 percent or more, and to research in Maryland finding that speed cameras reduced the likelihood of serious injury by 39 percent.
That second figure is the one that ultimately settles the argument. A reduction in speeding is an intermediate measure. Injuries and deaths are the outcome measure.
Laura Friedman, who wrote California’s speed camera law while in the state assembly and now represents part of the Los Angeles area in Congress, says the legislation was built with safeguards specifically to keep it a traffic-calming measure rather than a revenue mechanism.
Where the money goes
This is the most concrete response to the revenue objection, and the design feature most worth exporting.
Under AB 645, citation revenue can be reinvested in traffic calming infrastructure on the same corridors: speed tables, protected intersections and physical improvements.
The structure does two things at once. It ties the money to the road that generated it rather than to a general municipal fund. And it makes the effect durable, because physical design keeps working even when the perception of enforcement fades.
Oakland also operates fine reductions for low-income drivers, cutting a $50 citation to $25 or $10.
The transferable lessons
Automated speed enforcement is already widespread in many countries, including Türkiye. The question is not whether to install cameras.
Three elements of the California approach travel well:
Baseline measurement. What makes the SFMTA result credible is not the number of cameras but the pneumatic-tube baseline taken at the same sites beforehand. Without it, no claim about a programme’s effect can be evaluated.
Repeat-offender tracking. Oakland’s finding that two thirds of warnings went to repeat recipients is what turns an enforcement system from a citation machine into a behaviour-change instrument.
Hypothecated revenue. Directing fine income into infrastructure on the corridor where the violation occurred strengthens public legitimacy and makes the safety gain permanent.
The effect of cameras is not seriously in dispute. What is disputed is how that effect is measured and where the money goes. California has now put a written answer to both.
Sources
- SFMTA speed safety camera programme initial evaluation, March 2026; US DOT ITS Knowledge Resources record 2026-B02033
- Oakland Department of Transportation, Speed Safety Cameras Initial Report (14 January to 21 February 2026) and programme pages, oaklandca.gov
- The Oaklandside, reporting on the Oakland camera programme, January and May 2026
- KQED and KTVU, programme launch and initial data coverage, January and March 2026
- NPR, As speed cameras spread to more states, so does the backlash from drivers, Joel Rose, 16 August 2026
- California Assembly Bill 645
Reported by the roadsafetynews.com editorial desk. The 72 percent figure is San Francisco’s; Oakland has not yet published before-and-after speed data. Analysis is the editorial desk’s own.
