Loading...
HomeMy WebLinkAbout06/16/2026 04. Distributed at the Meetinged at Meeting: tem# Good evening city council, We have spent several months asking Council & YPD to remove the Flock Safety surveillance cameras, to end the contracts, & to invest those funds into much needed community programs. You have refused to do so. You have deliberately kept the community in the dark about these cameras, what they do, & what happens to their data. You have not allowed the community the opportunity to weigh in, have their say, or vote. You have allowed & perpetuated the misleading of Yakima Community members through your inaction & actions. You have asked for alternative solutions, which have been provided, and which you then ignored or dismissed. You have lied by omission & continue to do so. You have done no apparent research of your own, even minimally. You have taken the uninformed word of police chiefs Boyle & Soptich as facts without interrogation of their sales pitches or their bias & conflict of interest. You have not consulted with any outside digital security or constitutional legal experts. You have allowed & supported the widespread unconstitutional collection of our data. You have allowed that unconstitutionally collected data to be retained & shared by Flock however they choose with no regard for the harm that causes. You have not read the contracts. You have not read the laws. You have used ARPA funds without regard for or minimal review of the funding the treasury's guidelines, specifically relating to the provisions on how to use the funds, how to procure vendors including rules on any disbarred or suspended vendors, who retains ownership of the capital asset purchase, and conflicts of interest. You have put Yakima at risk for more lawsuits which the City can't afford. You have opened yourselves up to extensive litigation which could involve a class action, could involve multiple states, including federal courts, would involve constitutional issues, with potential requirements for massive amounts of discovery documentation including your decisions and the basis for them. Your failures here are widespread & dangerous. You are not erving this community & have proven untrustworthy to do so. SOME BASIC INF© ABOUT AN ALPR An ALPR is a motion -detecting surveillance camera. "Recording activity over a specific area with a Flock camera is possible, but Flock cameras have an intentionally narrow field of view. Our motion detection works up to 75 feet, and the field of view at that distance is —20 feet wide (about 1.5 lanes of traffic). The infrared technology that we use to capture license plates illuminates reflective subject matter, so anything that is not reflective (license plate) or illuminated (light pattern of the vehicle) will be completely dark. (https://www.flocksafety.com/faq) How many photos does it take at a time? "Flock Safety cameras capture somewhere in the range of 6-12 images of a vehicle per session. When a car drives by, the camera will rapidly capture it entering and exiting the camera frame. Our engineering team has the ability to configure those settings based on the camera location and how much traffic it is expected to capture. This ensures we can provide customers with the evidence they need based on the location." (https://www.flocksafety.com/faq) These cameras capture everything in the scene, including people, their clothing, their skin color, their shape, any equally identifying info that are identified in cars. What Flock says: "Regular security cameras record videos focusing on people and the overall context. License plate recognition cameras take still photos to focus solely on a vehicle and ensure a clear view of the license plate. Flock Safety takes the traditional LPR camera to the next level by using machine learning to capture clear, accurate license plate infoiuiation and organize it based on a vehicle's physical description and unique features. Flock Safety cameras apply machine learning to the footage they capture to identify important details that traditional LPR cameras overlook. Besides just a license plate number, Flock Safety captures and organizes vehicles into categories based on: • Make • Model ▪ Color • Resident or non-resident vehicle ▪ Timestamp • Type of plate (standard v. temporary) • Damage or alterations (Le. broken taillight, after -market wheels) Flock. Safety LPR cameras go beyond capturing license plates. Unlike traditional license plate readers, the unique Vehicle FingerprintTM technology accelerates investigations by allowing users to filter their search based on the vehicle's specific characteristics, including body type, make, color, and more. Flock Safety is able to capture vehicles traveling up to 100 MPH, day and night, and up to 75 ft away. That's a big part of how we help solve and reduce crime across the country." "What the Patent Says A review of Flock Safety's patent filings reveals language describing backend systems capable of analyzing visual data in ways that go beyond simple license plate recognition. While the cameras Understanding Flock S Privacy Feb 28 ; Updated: May 27 e s Claims and Implications for Flock Safety's Core Marketing Claim Flock Safety has built its public reputation on a simple and powerful claim: Their cameras do not use facial recognition technology. In a time when biometric surveillance is one of the most controversial technologies in America, that message matters. Many cities and states have passed laws restricting or outright banning facial recognition use by law enforcement. Public resistance is strong. So promising "no facial recognition" helps reduce fear, ease procurement, and win over communities. Flock markets its system primarily as an automated license plate reader (ALPR) network — focused on vehicles, not people. But here's where things get complicated. What the Patent Says A review of Flock Safety's patent filings reveals language describing backend systems capable of analyzing visual data in ways that go beyond simple license plate recognition. While the cameras themselves may not actively perform facial recognition at the edge device level, the patent documentation appears to describe: • Image processing pipelines • Object detection capabilities • Data aggregation systems • Searchable image databases • Pattern analysis across stored footage • Potential biometric or identity -based correlation features In other words, while the front-end marketing emphasizes "no facial recognition," the backend system architecture described in patent filings appears technically capable of supporting it. That distinction matters. Edge Device vs. Backend Capability There is an important technical difference between: 1. What a camera does today 2. What a backend system is designed to enable 3. What could be activated via software updates later Modern surveillance systems are modular. Hardware may collect data. Cloud infrastructure may process, store, and analyze it. Al capabilities can be added or upgraded remotely. If a patent describes backend architecture capable of biometric identification or facial vector analysis — even if not currently activated — it raises a legitimate public policy question: Is the "no facial recognition" claim about present use... or permanent technical limitation? Those are not the same thing. Why This Matters for Public Trust Cities across the United States — from small towns to major metro areas — have adopted Flock Safety systems. Municipal decision -makers often rely on vendor assurances when evaluating privacy risk. If the public is told: "This system cannot perform facial. recognition." But the patent language suggests the backend is architected in a way that could enable it, then critics argue that the marketing may be overly narrow or incomplete. Public trust depends on clarity. When companies use highly specific wording "we don't do facial recognition" — without addressing whether the system is technically capable of it, it can feel like semantic positioning rather than transparency. Is This Dishonesty — or Just Legal Strategy? To be fair, companies routinely file patents for capabilities broader than their currently deployed features. A patent protects possibility, not necessarily implementation. It is entirely possible that: • Flock Safety has no intention of activating facial recognition. • The patent language is defensive intellectual property strategy. • The company is complying with current law and policy in all deployments. However, when a company markets heavily around a privacy -sensitive claim, and internal technical documents describe systems that could theoretically contradict that claim, scrutiny is reasonable. The issue is less about whether facial recognition is currently active — and more about transparency regarding architectural capability. The Broader ALPR and Surveillance Debate Automated license plate readers already create large-scale vehicle movement databases. When paired with: • Cloud storage • Cross -agency data sharing • Al analytics • Long-term retention policies ... the privacy implications become significant. Adding even the potential for biometric expansion increases those concerns. Communities deserve clarity about: What the system does today What it is technically capable of What contractual safeguards prevent expansion Whether future updates could change functionality Key Questions That. Deserve Answers If you are a policymaker, journalist, or concerned citizen, consider asking: 1. Does the backend system include facial vector extraction capabilities? 2. Is facial recognition technically disabled — or architecturally impossible? 3. Are there contractual restrictions preventing future activation? 4. What audit mechanisms exist? 5. How is stored image data indexed and searchable? Transparency builds trust. Ambiguity erodes it. The Importance of Community Engagement Engaging with the community is crucial for understanding the implications of surveillance technologies. Residents should be informed about how these systems work and what data they collect. Public forums can serve as platforms for discussion, allowing residents to voice their concerns and ask questions. The Role of Local Policymakers Local policymakers play a vital role in regulating surveillance technologies. They must ensure that any implementation of systems like Flock Safety's aligns with community values and privacy rights. Policymakers should advocate for clear guidelines and transparency from technology providers. The Need for Ongoing Oversight Ongoing oversight is essential to ensure that surveillance systems do not overreach. Regular audits and assessments can help maintain accountability. Communities should demand that their local governments implement these measures to protect citizen privacy. Final Thoughts: Marketing vs. Architecture Flock Safety has successfully positioned itself as a privacy -conscious alternative in the surveillance technology market. But patents exist to protect technical capability — and sometimes those capabilities are broader than public -facing messaging suggests. Whether this represents dishonesty, legal caution, or future -proof engineering depends on interpretation. But one thing is clear: When companies operate at the intersection of AI, law enforcement, and civil liberties, precision in language matters — and so does architectural transparency. In conclusion, understanding the nuances of surveillance technology is essential. As a community, you must stay informed and engaged to ensure that privacy rights are respected. The conversation around surveillance technologies should be ongoing, fostering a culture of transparency and accountability. h s://www.deflockbcs.corni ost/flock-safet -sa s-its-cameras-don-t-do-facial-reco nition-so-wh does -its -patent -describe -it Flock says its cameras don't track people. Its own training videos say otherwise. Company's newest devices use AI to follow pedestrians, while protesters are mapping nearly 90,000 surveillance cameras nationwide. Flock's own training videos show police using the system to track suspects "from location to location to location." By Brendan Keefe Published: Jun. 15, 2026.at 10:33 AM PDT Flock Safety says its cameras don't track people. But the company's own training videos show police using the system to track suspects "from location to location to location." And some of its cameras are designed to follow people as they walk. "And they quite literally use AI to zoom in and follow you around whether you're a person of interest or not," said Benn Jordan, a security researcher who found dozens of Flock cameras streaming openly online. This is the second report in an ongoing investigation into Flock Safety, a private Atlanta - based company that operates a network of surveillance cameras used by thousands of police departments nationwide. Tracking the trackers Inside a coffee shop in Boulder, Colorado, Will Freeman runs the largest police camera tracking system in the nation. It's called DeFlock. The app shows where Flock cameras are located and which direction they're pointed. "How many cameras have you mapped?" our national investigator Brendan Keefe asked. "88,970 cameras at the moment," Freeman said. Freeman said he created the app to show people how invasive the surveillance network has become. Flock license plate readers take a time -stamped photo of every vehicle that passes. Police can search historical travel information for any car for 30 days without a warrant. "If you were to build a graph of that sheet and plot it to a map, now it's as if you've had a GPS on your car for an entire month," Jordan said. The tracking debate Josh Thomas, Flock Safety's chief communications officer, said the company helps solve 700,000 crimes a year. "Reported crimes in America, yeah," Thomas said. "So there's a lot more that probably are being helped that aren't being reported. But yes, we have about 700,000 crimes that we are aware of that are being solved, in part with evidence that's derived from a Flock Safety device." But Thomas disputed the characterization that the system tracks people. Thomas said the system captures fixed points in time, not continuous tracking. "Tracking the whole of your movements is very different than knowing that you pass by one specific camera, 24 specific cameras," Thomas said. "Those are different things." Flock's own training videos But Flock's own webinars describe the system's ability to track vehicles and people. "The example of tracking that vehicle from location to location to location," a Flock webinar instructor said. "And you're able to track your suspect's movements," another webinar showed. In one training video, a police officer described using Flock cameras to follow a suspect across state lines. "And we were able to track him all the way over to another state, in Kentucky," the officer said. Condor cameras track pedestrians Flock's Condor cameras are different from its license plate readers. The Condor is a pan -tilt - zoom camera that uses artificial intelligence to detect and follow people. "You're saying Flock does not track people, correct?" National investigator Brendan Keefe asked Thomas. "Track people? That's right," Thomas said. "And the whole of their movements, like you asked, we're not tracking people." "But what do the Condor cameras do? They literally are designed to track a person, aren't they?" Keefe asked. "Not to track a person, no," Thomas said. But when Keefe walked in front of a Condor camera on a public trail, the camera followed his movements in real time, panning and tilting to keep him in frame. A Flock training webinar describes the feature. "To alert you when it sees human movement," the webinar instructor said. "And it will actually acquire and track that person as they walk, or people." Thomas said the camera has an object detection feature called Guardian Mode. "There's an object detection feature that you can set up, it's called Guardian Mode, that allows the camera to just know if there's movement from a person in the screen," Thomas said. "It is not tracking where you're going to go, where you're going to leave, or you're going to start and stop." Jordan said the Condor cameras appeared shortly after Flock said its cameras only record license plates. "These started popping up weeks after hearing Flock say, `These only record license plates. They don't take pictures of people," Jordan said. The Condor is a pan -tilt -zoom camera that uses A.I. to detect and folio w people. (InvestigateTV) Cameras streaming without passwords Jordan found the same Condor camera streaming openly online with no password required. He published the video on YouTube along 404 Media. "At one point, I think we had found over 70 cameras," Jordan said. Jordan said he was able to watch people in real time. "I watched a man leave his house in the morning. I watched a woman jogging alone on a forest trail in Georgia," Jordan said. Jordan was able to read a statement from Flock's CEO about the company never being hacked in front of one of the exposed cameras. He downloaded the footage later over the internet, directly from the camera. Thomas said the exposure was an accident caused by Verizon. "This wasn't the dark web," Keefe said. "This was openly streaming to anyone who could go on a specific search engine and scan open ports. Was that an accident?" "Yes," Thomas said. "Verizon sent us the wrong SIM cards that had a public IP address. We have a lot of sensors in the world that don't have that. Roughly 60-70 devices had a public - facing IP address. The second we learned about it, we fixed it." Verizon did not respond to a request for comment. Officers arrested for misuse Police officers nationwide have been arrested for using Flock cameras to stalk former partners and love interests. "What if a police officer uses Flock cameras to stalk their ex? What if an abusive partner uses it to stalk their ex -girlfriend and her new boyfriend? It's already happened/' Jordan said. Thomas said misuse is a horrible use of the technology. "Tools aren't designed to do that," Thomas said. "This is why we built an audit log, and we work with the appropriate democratically elected governing bodies to hold those people accountable." Freeman said abuse is inevitable. "You have to consider human nature when you implement things like this," Freeman said. "No one's perfect. People are going to abuse it." Routing around cameras Back in the Boulder coffee shop, Freeman demonstrated another feature of DeFlock. "Direct route, four cameras," Freeman said, showing a 1.7-mile route. "And that's just to go 1.7 miles." Will Freeman shows how the app DeFlock suggests alternate routes to avoid Flock cameras(InvestigateTV) DeFlock allows users to plot routes that avoid cameras. The app showed an alternate route that took nearly three times as long but avoided all cameras. "Yeah, it goes from five minutes to 14 minutes, but it avoids all the cameras," Freeman said. The debate over public safety Thomas said the technology helps prevent mass killings. "There's an epidemic, as we all know, of mass killings in America," Thomas said. "And if there's an opportunity to stop that, I want to be on the side of the people who are fighting to stop those things, not the people who are fighting to stop the proliferation of technology to stop crime." "Actual killers have been caught with this technology," Keefe said to Freeman. "Would you want to take that away from the police?" "Considering everything, I would want to take that away from the police, at least in the current state that it's in," Freeman said. Freeman said the problem is that the system logs everyone, not just suspects. "A feature that you can't turn off is that it's logging everyone else," Freeman said. "So if they can get rid of that 'logging everyone else' part, 1 think it would be a lot better." Freeman said he will continue tracking the trackers because no one is watching the watchers. More than two dozen cities recently cancelled their Flock contracts, including Denver, amid concerns over who has access to the company's camera data. Copyright 2026 Gray Media Group, Inc. All Rights Reserved. https://www.cleveland19.com/2026/06/15/flock-says-its-cametas-dont-track-people-its-own- training-videos-say-otherwise/ Photos of cameras seen downtown Yakima June 14u' and are still there June 15t at least: ;,44191500,1AttAsin, C;;;;710,01,00 1,p4NtOtgOlrWe' k$4.& 101,1414,, Virpirgep