Flock Safety is marketing its cameras to neighborhood groups and law enforcement with the promise to “reduce crime in your community by up to 70%.” Instead of selling these cameras, it leases them for about $2,500 per month. The cameras can be powered with solar panels and transmit data via a cellular modem so no extra wiring is needed. Note that the version we dissected is an older version and the modem is no longer being sold. Although the cameras capture images that include the license plates of vehicles, they are triggered by a motion detection sensor and will likely capture an image of anything that triggers the motion detector, including people, bicycles, etc.
Here’s the list of component parts that we were able to identify:
Cellular: NimbeLink Skywire LTE modem – FCC ID N7NHL7648
There are 32 agencies that submit license plate reader data to the Northern California Regional Intelligence Center (NCRIC), the regional fusion center for Northern California. However, many more agencies are able to access that data once it has been sent to the NCRIC.
Other agencies accessing NCRIC’s license plate reader database are to be expected, such as the police departments from the Bay Area’s three largest cities: San Francisco, San Jose, and Oakland. After the three biggest users of the NCRIC license plate reader database, the users vary from Fremont, with an estimated population of 240,000, to cities like Dixon and Atherton, with populations of less than 20,000.
Piedmont, with more than 39 automated license plate readers situated at the border with Oakland, has been one of the three largest submitters of license plate reader data to NCRIC, with 22 million images of vehicles of license plates and vehicles annually. However, during the period from February 21 to March 21, 2021, it didn’t use the NCRIC license plate reader database even once. Vallejo, which submits more than 15 million images of vehicles of license plates and vehicles to NCRIC annually, queried NCRIC’s database just four times in this one month period. Berkeley, which doesn’t submit any license plate reader data to NCRIC, queried the database 61 times during this one month period.
According to records for October 2021, Piedmont queried the NCRIC license plate reader database just once while Vallejo didn’t query the database at all.
A table showing the agencies and the number of queries to the NCRIC license plate reader database for 2/21/21 to 3/21/21 is shown below. Note that queries by a District Attorney’s office may be included for some of the agencies listed, like the San Mateo, Alameda and Solano County Sheriffs. In addition, Danville, which is listed below, is contracting out its policing to the Contra Costa County Sheriff.
Although some law enforcement agencies publish data on the number of license plates photographed by license plate readers, few provide information on why police search for license plates. According to a document released by the Northern California Regional Intelligence Center, (NCRIC) the most common search in October 2021 was related to “Locate Stolen, Wanted, or Suspect Vehicles.”
When law enforcement agencies search NCRIC’s license plate database, they select one of six reasons for why they are searching for a license plate in the database:
Locate Stolen, Wanted, or Suspect Vehicles
Locate Suspect(s) of Criminal Investigation or Arrest Warrant
Locate Witnesses or Victims of Violent Crime
Locate Missing Children and Elderly individuals (Amber / Silver Alerts)
Protect Critical Infrastructure
Protect the Public during Special Events / Situational Awareness
For October 2021, 1015 or 67% of the searches were for stolen, wanted or suspect vehicles. The next was common search was to locate suspects with 442 searches or 29%.
The least frequent searches were to locate witnesses or victims with 33 searches (2%), locating people for amber or silver alerts with 12 searches (0.8%), protecting critical infrastructure with 8 searches (0.5%) and protecting the public/situation awareness with 4 searches (0.25%).
San Ramon, a small city in Contra Costa County with a very low crime rate, approved spending more than $1.2 million to blanket the city in automated license plate readers (ALPRs) and surveillance cameras at its April 28, 2020, meeting.
The amount includes 12 stationary ALPRs at five locations from Vigilant Solutions, 35 ALPRs from Flock Safety and 46 Avigilon surveillance cameras. Eight police vehicles are already equipped with Vigilant ALPRs and Flock Safety cameras are already installed at the intersection of Camino Ramon and Crow Canyon and the northeast intersection of Crow Canyon Boulevard and Crow Canyon Place.
According to a presentation by San Ramon Police Lieutenant Cary Goldberg on April 28, 2020, the surveillance cameras would offer “coverage on every route of ingress/egress,” “184 viewing angles throughout out the city” and vehicles would likely pass “multiple cameras when travelling [sic] in San Ramon.”
Goldberg said that data from Flock Safety cameras can be searched for animals and “If the camera detects an animal in the car, it will bring up that particular picture of the vehicle if it, in fact, captured it.” The data can also be searched for bicycles and people. “If it detects a person in a car, it’s going to show you an image of that person.” Goldberg said the locations were chosen so “we can capture all our routes of ingress, coupled with the situational [surveillance] cameras so we have a virtual fence around the city.”
During the presentation, Goldberg showed an image captured by a Flock Safety ALPR at Crow Canyon Place at Crow Canyon Road of a Honda Odyssey van with its California license plate clearly visible. Another image from the same location at night provided the license plate of a Ford pickup and the fact that the license plate had been seen 13 of the last 26 days (indicating that it was likely driven by a resident of San Ramon). We have redacted the license plate numbers in the presentation produced in response to our public records request.
Each of the Avigilon surveillance cameras actually has four cameras to provide 360-degree coverage. In addition to recording video 24×7, “The officers also have the ability to watch this live. Let’s say if we had an event, we had a protest, whatever the case is, the officers would have the ability to log in to one of these through the software and watch these live, watch any number of these cameras live.” According to Goldberg, a vehicle can be tagged and the software can display any video captured of that vehicle without searching individual camera footage. The software can also generate alerts for unusual motion detection, such as a car crash, according to Goldberg.
Proposed locations for the Vigilant Solutions ALPRs include Dougherty Road, the intersection of Camino Tassajara and Windermere Parkway, Crow Canyon Road, and the intersection of Alcosta Blvd and San Ramon Rd. [Note: The slide describing the Vigilant ALPRs showed an image of 3M ALPRs.]
The presentation to the San Ramon City Council said that the data collected by Vigilant Systems ALPRs would be retained for two years. Data from the Flock Safety ALPRs would be retained for 30 days. Data collected by the Avigilon surveillance cameras would be retained for 30 days. The San Ramon Police Department policy for license plate readers only states that the data will be retained for a minimum of one year. The police didn’t explain the difference in retention periods or the conflict with the existing policy and the City Council didn’t ask. When we asked Lt. Goldberg about why images from the surveillance cameras and Flock license plate readers were retained for 30 days while data captured by Vigilant Solutions license plate readers is retained for two years, he responded (but didn’t explain), “we have continuously maintained a 2 year retention period for the images uploaded into the LEARN [Vigilant Solutions] database.”
San Ramon Councilmember Phil O’Loane raised the only objection, based on the cost of the project, which was nearly double what he was expecting. O’Loane said, “My understanding of where we were starting was stationary license plate readers at key intersections and such and we’ve gone to something between zero and the City of London” noting that San Ramon has one of the lowest crime rates in the state. “I thought we were going to spend six or seven hundred thousand dollars and we’re doubling that. There’s been significant scope creep on this.” “We’ve got the same crime rate we’ve had since I was on the Council on average.” In fact, San Ramon’s crime rate is not only low but has declined significantly in the past 30 years.
The figure below is based on data reported by the San Ramon Police Department to the FBI and represents the annual crime rate per 100,000 people from 1985 to 2018.
According to a $60,000 grant the Novato Police Department recently received from the Bay Area Urban Areas Security Initiative, new automated license plate readers (ALPRs) will be installed on Highway 37 to track terrorists.
However, no suspected terrorists have been tracked or identified with Novato’s existing ALPRs, which have been in use since 2012. We asked Novato City Manager Adam McGill for information on how many suspected terrorists had been detected or tracked with Novato’s existing ALPRs. McGill forwarded the request to Novato Police Lieutenant Christopher Jacob, who responded, “According to our records, we have not detected any hot plates related to or associated with terrorism.”
The grant application, filed in September 2019, claims that “License plates captured and identified by ALPR can alert authorities to wanted vehicles associated with suspected terrorists…”
“This project will install a 4 camera ALPR system on a preexisting CalTrans traffic operations pole located along the southern edge of US Hwy 37. This route connects US Hwy 101 to US Hwy 80 and serves as the primary route of travel into and between Marin, Sonoma and Solano counties. This location, according to recent CalTrans vehicle census information, has the potential of capturing and identifying no less than 12.7 million license plates a year.”
According to the FY 2016/2017 Proposed Budget and the 2016 Novato Police Annual Report, the existing ALPRs captured 1,195,272 images of vehicles and license plates in 2015 and 1,204,854 in 2016. With an estimated 2018 population of 55,655, that’s more than 26 license plate reads each year for each Novato resident. With an additional 12.7 million license plate reads per year, that would increase to more than 249 license plate reader each year per Novato resident. According to the Novato Police Department policy for ALPRs, the data collected is stored for a minimum of one year.
The 2012 Novato Police Annual Report states that the department received a $20,000 Homeland Security grant for mobile ALPRs to “identify, track and report wanted terrorist suspects and locate other wanted suspects identified by the Department of Justice.” ALPRs were installed on five police vehicles in 2012 and by 2014, an additional police vehicle had ALPRs installed. A mobile ALPR trailer funded by a $36,583.75 grant from the California State Homeland Security Grant Program was added in 2014.
The Vallejo Police Department cited ferry terrorism as part of its $30,000 grant application for license plate readers, calling the ferry “a viable threat for a terrorist act.”
According to the grant application, the “Grant Funds will be utilized to establish an ALPR [Automated License Plate Reader] camera system for eastbound and westbound traffic lanes approaching the Ferry Terminal, Ferry Terminal Parking Structure and waterfront.”
Vallejo Detective Lieutenant Fabian Rodriguez was listed as the applicant for the grant for the project, which was expected to begin on November 1, 2020 and completed by May 1, 2021.
The regional benefit was described as:
ALPR Cameras to Strengthen Information Sharing and Collaboration of data from Vallejo Ferry Terminal. The Ferry Terminal provides daily commuter transportation from Solano County to San Francisco, with commuters from Napa utilizing the ferry for daily transportation to the Bay Area.
The terrorism nexus was described as:
With thousands of commuters utilizing the Vallejo Ferry system each day to travel to San Francisco, the Ferry is a viable threat for a terrorist act. Installing an ALPR camera system in the area of the Ferry Terminal will assist with identifying potentially wanted terrorist suspect vehicles.
On March 12, 2020, the Bay Area UASI Approval Authority approved a $30,000 grant to the Vallejo Police Department for its “Vallejo PD ALPR” project. When we requested a copy of the grant application, the Vallejo Police Department responded that it has “no responsive documents.” The grant application was disclosed pursuant to a public records request to the Bay Area UASI.
According to information released in response to a public records request, the Northern California Regional Intelligence Center collected more than 79.2 million license plate reader records from 32 local law enforcement agencies from June 2018 to May 2019.
Piedmont, a city with about 11,000 people, sent the most data to NCRIC, with more than 22.4 million license plate reader records, which include license plate numbers and photographs of vehicles and their surroundings. Piedmont has more than 30 stationary license plate readers that capture nearly all traffic coming into and out of Piedmont.
In calendar year 2018, Piedmont collected 21.6 million records from its license plate readers and reported 8,120 hits, when a record from a license plate reader matched a list of license plates that includes stolen vehicles, stolen license plates, wanted persons, etc. Using that information, 99.96% of the data collected by Piedmont’s license plate readers is from people who are not suspected of are charged with any crime. That closely matches a similar analysis in 2014 that showed 99.97% of the data collected from Piedmont’s license plate readers did not generate a hit.
Fremont, a city with about 230,000 people, sent 17.7 million records to NCRIC and Vallejo, a city of 120,000 people, sent 15.8 million records to NCRIC.
Other agencies sending more than a million records to the NCRIC each year include the Central Marin Police Authority (Larkspur, San Anselmo and Corte Madera), Daly City, San Francisco, Modesto, Alameda County Sheriff, San Leandro, and South San Francisco.
Other law enforcement agencies use license plate readers from Vigilant Solutions, a private company that collects data from law enforcement agencies and private companies. Data from 2017 indicates that Bay Area agencies in Danville sent 33.4 million license plate records to Vigilant, Pittsburg sent 31.4 million records, Brentwood sent 12.9 million records, and Alameda and Novato each sent 1.6 million records.
Of the 28 jurisdictions sending license plate reader data to NCRIC, Fremont collects the third-highest amount of data, behind Vallejo (21.7 million) and Piedmont (21.3 million). NCRIC is a regional fusion center that provides a license plate database where agencies can send their collected data. Other agencies, including the IRS, Department of Homeland Security, FBI, National Park Service, California Department of Insurance and others can access this database and search for license plate data.
The license plate reader on westbound Stevenson Boulevard near the entrance to northbound Interstate 880 generates the most data, collecting an average of 14,736 license plates and photographs each day during October 2017.
Although similar statistics are not available for Fremont, analysis of license plate reader data from other jurisdictions has found that 99.7% of the data collected by license plate readers is of people not suspected of or charged with any crime. For Fremont, that means 14,519,834 of the 14,563,525 photos and license plates collected during an 11-month period was of innocent people.
The Center for Human Rights and Privacy recently discovered that the Alameda County Sheriff installed a license plate reader at the entrance to Highland Hospital’s emergency room. The information was included in a list of 29 agencies that submit license plate reader to the Northern California Regional Intelligence Center (NCRIC), a local joint fusion center.
However, when the East Bay Times asked about its license plate readers in February 2017, the Alameda County Sheriff’s Office reported that it had about six license plate readers, mounted to patrol cars and didn’t mention the license plate reader at Highland Hospital.
Although the license plate reader was installed as part of an Alameda County General Services Agency project, it appears that the Alameda County Sheriff configured the license plate reader to send its images and data to NCRIC. The Alameda County Sheriff signed a Memorandum of Understanding with NCRIC in October 2014 regarding license plate reader data. According to NCRIC, the Alameda County General Services Agency did not have a MOU with NCRIC as of November 20, 2017.
The license plate reader has two cameras to capture images of vehicles and license plates as vehicles enter the emergency room drop-off area of Highland Hospital from E. 31st Street.
According to information from NCRIC, the license plate reader captured data from 293,148 vehicles from December 2016 to October 2017. Earlier data was not available. Once the information is submitted to NCRIC, it is made available to dozens of agencies, including Immigration and Customs Enforcement (ICE) and Homeland Security Investigations (HSI), which can use it to identify undocumented immigrants for deportation.
With a nod to Terminator’s self-aware computer network, the Solano County Sheriff in Northern California has applied for and received partial funding for Project Skynet, a network of surveillance cameras and automated license plate readers (ALPRs). Project Skynet would install 160 ALPRs and 124 surveillance cameras at 66 locations throughout Solano County. In its response to a request for public records, Daniel Wolk, Deputy County Counsel for Solano County noted, “this is at the proposal stage and specifics, including camera locations, have not been decided upon.”
Map of Phase I Camera/ALPR Locations
Like the ring of stationary ALPRs around the City of Piedmont, the goal of the project is to capture images and video of all vehicles entering and exiting Solano County. According to a Microsoft PowerPoint presentation prepared by the Solano County Sheriff’s Office, the surveillance cameras would be capable of pan-tilt-zoom and data from the ALPRs and cameras would allow “real time information or years of past history.” The UASI grant application states, “In addition to the direct connection this project has with observing and locating terrorists, it also allows law enforcement to monitor human trafficking. I-80 is a major artery for drugs and human trafficking which are major funding sources for terrorism, from the west coast to the east.”
As with all grant applications to the Urban Areas Security Initiative, the project must have a nexus to terrorism. In its proposal for a Homeland Security Grant, the Solano County Sheriff writes, “We will be able to…locate persons and vehicles associated with terrorist threats, disrupt terrorist financing…” However, given that terrorism is extremely rare, this surveillance network is more likely to be used against pedestrian suspected criminal activity.
The PowerPoint document states, “There are 36 (666) locations, many of which are large freeways and near impossible to watch at once.” “666” is a police code for a county-wide emergency or be-on-the-lookout (BOLO). However, the grant application and approval documents cite 35 locations.
The estimated $2 million cost for Project Skynet is broken down into phases with an estimated cost of $535,000 for Phase I, $568,000 for Phase II, and $895,000 for Phase III. Data from the ALPRs would be stored at the Northern California Regional Intelligence Center (NCRIC), the local joint fusion center that shares license plate reader data with dozens of local, state, and federal agencies.
The NCRIC stores license plate reader data for dozens of local law enforcement agencies and as of April 2015, stored 46.5 million records, which includes license plates and photographs of vehicles, including the occupants and surrounding area with geolocation data.