
Berkeley has installed just over 60% of its planned automated license plate reader (ALPR) network. Police have used the network to close a half-dozen cases — most involving stolen cars, one involving a suspected shooter — and arrest eight people over the first month of using it.
But the Berkeley Police Department says it’s too early to offer a more comprehensive evaluation of the network, according to a report to the Police Accountability Board Wednesday.
The City Council approved a two-year, $425,000 trial period with Flock Safety in October 2023 to install 52 of the readers, which capture still images of passing vehicles. The 32 cameras the city planned to install on its own streets are already in place and in use. The city has not yet installed the remaining 20, mostly along Ashby and San Pablo avenues, in rights-of-way owned by Caltrans.
A select few BPD personnel have begun using the network for investigations, and are working to determine how best to triage the barrage of data it can provide.
The plate reader network compares plates of passing cars to local, statewide and national databases of stolen cars and cars connected to other serious crimes, and sends notifications to law enforcement agencies in real time. Police have said they hope the technology will help them curb auto theft and other serious crimes, but skeptics have raised privacy concerns and other worries, such as if BPD could be compelled to share data with federal immigration agents.
City police entered a roughly month-long “introductory period” in October, during which a limited number of officers, detectives and supervisors began using the network, City Manager Paul Buddenhagen announced in an Oct. 15 memo. Police said that period is still in progress, and there is not yet an estimate for when the network will be fully operational and all users fully trained.
This photo shows an example of a stationary automated license plate reader. Berkeley has installed 32 and intends to install another 20. Credit: Flock SafetyOfficers can access the network by mobile devices or computers installed in police cars, and must review images captured by the network to ensure the system did not misread a license plate or misinterpret a physical characteristic of the car in question in order to avoid any false hits, Rafferty said.
“A Flock hit or an alert, that alone does not justify an enforcement stop,” Berkeley police Sgt. Darrin Rafferty told the board Wednesday. “That is merely an alert letting you know that there’s a little more work that you’ve got to do before you make any enforcement stop.”
While the cameras are fairly accurate at capturing photos in daylight, their precision at night is not as good, generally only capturing images of license plates — thanks to their reflective coatings with which many states print their plates — rather than entire cars, Rafferty said.
Officers, detectives and supervisors can customize which alerts they receive. Rafferty said most of the current network users have turned off their alerts for stolen license plates because the network was sending roughly 90 daily alerts for them, “and there’s additional investigation that you need to do if you would want to take any enforcement action on that.” The tryout period team has focused instead on alerts for stolen vehicles or vehicles connected to felonies, amber alerts or “hot lists,” which law enforcement agencies compile based on active investigations.
To put the number of stolen plates in context, Berkeley police spokesperson Officer Byron White clarified that the network records tens of thousands of “unique” reads — a single vehicle, no matter how many readers spot it as it moves through Berkeley, counts as just one unique read — every day.
Critics have said the surveillance technology poses grave privacy concerns and endangers religious minorities, migrant residents and patients seeking reproductive healthcare and other treatment not available out of state.
“We do not permit the sharing of ALPR data for federal immigration enforcement, with ICE or Customs and Border Protection,” Rafferty said Wednesday. Whenever someone performs a search on BPD’s plate reader network, the interface compels them to acknowledge a disclaimer not to use Flock’s tech for immigration enforcement, he said.
Related to the department not sharing data with federal immigration informants, board member Kitty Calavita raised the issue of BPD possibly sharing or being compelled to share data with other states with strict bans on women’s reproductive rights. Rafferty said there was not yet a specific policy in place around that possibility.
Board member Alexander Mozes asked how using the readers might change how officers operate in a given shift, to which Deputy Chief Jen Tate said it was too soon to say.
Nor could BPD offer too many details on how effective the readers will be, in part because only a limited number of police personnel have even been trained on the network, Tate said. But, she said, in six cases in which BPD has relied on the readers over the last month, they have arrested eight people, and “a majority” of those cases involved stolen cars, she said.
BPD also partly relied on the readers to identify, track and arrest the San Francisco man they believe was firing off rounds near UC Berkeley on Oct. 26 and seize his arsenal of a dozen firearms. Police said the man, Jeffrey Darren Hue, was the shooter behind a “series of gunfire incidents” near the southern edge of the university campus, which witnesses and ShotSpotter gunfire sensors saw or heard in several different locations that morning.
An ALPR hit, for a car stolen from Berkeley in September, led to a police pursuit into Oakland on Oct. 16, ending in the arrests of two people.
Berkeley also plans on installing a dozen new fixed surveillance video cameras around the city by year’s end, adding to those already in use at several locations around the city, according to Buddenhagen’s October memo.
A list of locations for the readers is available on the city’s website.
"*" indicates required fields
Send a private note to the editors.*
See an error that needs correcting? Have a tip, question or suggestion? Drop us a line.

English (United States) ·