Back to All-In Podcast

Mass Surveillance, Police Misuse, and Who Controls Your Flock Cameras with Flock CEO, Garret Langley

All-In PodcastAugust 18, 202655m
Topics37
Introduction to Flock Safety and the Safety-Privacy Debate0:00Community Decision-Making Process for Camera Installation4:30Data Retention Policies and Effectiveness7:00Flock's Technical Limitations and Privacy Features11:02Communication Challenges and Public Response13:30Police Misuse Prevention and Audit Systems17:30Proposed Additional Safeguards22:31Multi-Approval Systems to Prevent Police Abuse23:31Challenges with Different Customer Sizes24:00Tailored Rules Based on Department Size24:32California Database Abuse Statistics25:01Evolution of Corporate Responsibility25:30Strong Defaults with Democratic Flexibility26:33Ring's Super Bowl Feature Controversy27:32Learning from Technology Concerns28:01AI and Police State Concerns28:30911 System AI Risks29:30Smarter AI Implementation Approaches30:30Pattern Recognition AI Limitations31:00Human-in-the-Loop Requirement31:31Cities Turning Off Flock Cameras33:01Consequences of Disabling Cameras34:02Austin, Texas Case Study35:01Competitors' Approach36:30ALPR Technology Capabilities38:01San Francisco Crime Reduction39:30Privilege and Security Access40:31Community Decision-Making43:00Personal Privacy Boundaries44:31PR Challenges and Internal Impact45:32Company Growth and Customer Retention46:59CEO Leadership and Anticipating Societal Impact47:31Media Appearances and Communication Strategy49:00Drone Technology and Rapid Response49:31Transparency Tools and Third-Party Auditing51:31Regulatory Landscape and Privacy Standards53:31Community Control and Decision-Making55:01
In a Nutshell

Flock CEO Garret Langley defends the company's license plate reading cameras as a community-controlled safety tool that has helped solve over a million crimes and locate 10,000 missing people, while emphasizing that cities—not the company—make final decisions on deployment and data retention periods ranging from 7-30 days. The company has implemented automated audit systems to detect police misuse after manual logs proved inadequate, requiring mandatory abuse detection features and developing multi-approval workflows to prevent stalking by the estimated 1% of problematic officers. Despite privacy concerns causing 60-70 cities to disable cameras, Flock maintains that communities benefit from choosing between privacy and safety rather than facing outright bans, and is expanding into drone technology with transparency portals for public oversight.

AI-Generated Notes

These notes were generated by AI and may contain inaccuracies.

Flock Safety was founded nine years ago with the mission to live in a safer neighborhood. The company began by selling license plate reading cameras to neighborhoods. The technology captures still images of vehicles, reads license plates, and has become more sophisticated over time to detect features like roof racks, bumper stickers, dents, and damage. In the past year, Flock has had a significant impact, solving over a million crimes and locating over 10,000 missing people, ranging from grandparents with dementia to children. A recent example involved a stolen car in Atlanta containing a 2-year-old child who was safely returned within an hour through the use of Flock cameras and Atlanta Police Department efforts.

The company operates in roughly 6,000 cities across the country. This widespread adoption has sparked debates about privacy versus safety, with city councils making final decisions on implementation.

Flock cameras are installed through a democratic process where city councils have the final decision. City councils meet every two weeks, and police chiefs, fire chiefs, or public works directors present concepts for approval. Flock representatives attend approximately 10,000 city council meetings annually to explain the product capabilities and limitations. Council meetings are recorded as public record, and the first council vote for Flock in Jersey Village, Texas, remains publicly accessible six years later.

Data retention periods significantly impact system effectiveness. Deleting data within one to two days reduces effectiveness compared to longer retention periods. Thirty days provides substantial effectiveness, while one year offers maximum effectiveness. Flock's default retention was 30 days but recently shifted to 7 days. City councils or states should democratically determine retention periods rather than the company dictating terms.

Approximately 90% of crimes can be solved within seven days of data retention. The remaining 10% includes complex cases such as homicides discovered days later through welfare checks or sexual assault cases where victims delay reporting due to shame or guilt. Most states have adopted regulations setting maximum retention limits, with New Jersey requiring five years and California and Washington establishing maximum limits.

Flock cameras do not perform facial recognition, capture video, look inside vehicles, or allow searches for people. This approach directly contrasts with competitors who offer these capabilities. The company positions itself as privacy-first compared to competitors who record full video and perform facial recognition.

Private communities often use multiple cameras recording everyone entering and exiting, and private property owners can legally record license plates of vehicles passing by. Flock emphasizes that communities using non-Flock systems may have more invasive capabilities.

Some individuals have vandalized Flock cameras, often influenced by social media content on TikTok and Instagram. One individual arrested in Ohio or Illinois faces 15-20 years in prison for camera vandalism. Flock initially referred to vandals as "terrorists," which was later acknowledged as a communication error. The company recognizes that some people hold valid privacy concerns and prefer the term "activists" as a better descriptor.

Some cities have removed Flock systems due to privacy concerns and backlash. The company accepts community decisions to discontinue use, though it considers reduced safety an unfortunate outcome when alternatives exist to limit data retention rather than complete removal.

Flock initially implemented basic audit logs, assuming this would prevent abuse. However, large agencies with thousands of officers making multiple daily searches made manual review of audit logs impractical, as supervisors could only spot-check a small fraction of searches.

Four months ago, Flock developed "audit assistance" - an automated tool that identifies abnormal usage patterns. Normal usage involves officers adding license plates to hot lists so entire departments can assist in locating suspects. Abuse patterns include repeated individual searches for the same vehicle without adding it to department-wide hot lists, suggesting personal stalking rather than legitimate investigations.

This system has identified multiple cases of police abuse. In Georgia, some police chiefs publicly announced firing nine officers for system abuse - the most severe offense under state law. Flock now requires audit assistance as a mandatory feature rather than optional default.

The company acknowledges that approximately 1% of police officers may engage in misconduct, which becomes significant in large departments like New York City's 30,000-40,000 officers or Los Angeles's 10,000-15,000 officers. The system is designed to identify and remove officers who abuse access privileges.

Additional safeguards could include requiring supervisor approval for searches, where officers must explain case justification to a sergeant, and a third person reviews sergeant approvals. This multi-level approval system could further prevent abuse while maintaining investigative effectiveness.

Flock is developing a system requiring two approvals for certain actions to prevent abuse by dirty cops. The implementation shows screenshots of how this would work and is on the whiteboard for curbing abuse.

Customers range from large departments with thousands of employees to small towns with 20 officers. The burden of multi-approval systems would be significant for smaller departments. A Texas police chief with 20 employees doesn't use the product daily and has only 2,000 residents, making the audit log very slim.

Different rules should apply based on city size - big cities have different rules than medium cities, which differ from small towns. This mirrors how becoming a police officer in New York City requires a longer academy than in small towns. Abuse is an industry-wide problem, not specific to Flock.

California had 7,000 reported abuses of their state database last year. The Washington Post covered this issue. Technology providers are now responsible for the worst behaviors of their customers, requiring ownership of privacy-security tradeoffs.

Previously, the approach was "it's a tool, do with it what you will," similar to how laptop, camera, and taser companies aren't held responsible for misuse. However, this technology is qualitatively different because Americans value privacy. The company's view of responsibility has evolved.

Elected officials rely on police chiefs, who rely on Flock to educate them. Responsibility falls on Flock to provide strong defaults that communities can pivot against while respecting the democratic process. As the lead candidate, the company takes the arrows for these decisions.

Jamie Simonoff's Ring cameras allow individuals to opt in to sharing data with police. During the Super Bowl, Ring released a feature to find lost dogs but had to turn it off due to concerns that the same technology could enable stalking.

Flock and Ring have learned to go slower in some cases after underestimating people's concerns about technology impacts. AI features are being developed more deliberately rather than launching feature after feature.

In a police state like China under CCP, facial recognition tracks people and monitors behavior like littering or raising voices. Americans see Flock and Ring on one end and the CCP's system on the other, questioning the distance between them.

When calling Delta, AI agents have low hallucination risk, but 911 systems are different. AI handling emergency calls could start talking about lunch instead of responding to a home break-in. Cities are leaning into AI for 911 due to budgetary and staffing issues, with some bad cases where AI got things wrong.

A smarter approach would have AI only pick up when phone lines are fully busy rather than replacing human operators. Flock doesn't handle 911 calls but sees this as a huge risk unfolding in the industry.

AI could identify abnormal patterns like a car slowly driving up to houses multiple times, but this raises concerns about predictive policing. Flock doesn't want to define suspicion, which is considered dangerous.

There's a valuable human element in public safety decisions affecting people's lives. AI products would require third-party attestation that systems work as expected, making officers more effective and objective rather than subjective.

Approximately 60-70 cities out of 6,000 (1%) have turned off cameras, often due to misinformation. One customer left after googling "ALPR facial recognition" and seeing a competitor's website, not realizing they were looking at the wrong company.

A Washington city turned cameras off, and the next weekend a little boy was shot with the car driving away. The neighboring city still had Flock cameras, found the vehicle, made an arrest, and provided justice for the family despite the tragic outcome.

Austin canceled Flock in 2025 but reconsidered after a shooting involving juveniles aged 15, 16, and 17. The teens committed crimes in Austin, drove to Manor where police used Flock cameras, and were arrested within 28 hours. The ACLU proposes 48-hour retention limits, while most cops want 30 days and communities typically set 7-14 days.

Some competitors run more recklessly without accountability tools or scrutiny. Flock has lost customers who want un-tethered systems without constraints, but the company doesn't want such customers anyway. Nearly 20 cities have turned cameras back on after realizing they were less safe.

ALPR systems capture not just license plates but also dents, car color, make and model. Most 911 callers don't remember license plates but remember unique features like damage on the right side or a missing hubcap. This technology helps start investigations rather than conclude them.

San Francisco has seen reduced crime and stolen vehicles. The "Flock effect" is one small part of a larger transformation including stronger elected officials, law enforcement, police chief, district attorney, and community demand for safety.

People with gates, guards, cameras, security systems, and personal security often don't understand why others need additional protection. Those who experience crime directly are more supportive of these measures, while wealthy individuals in areas like Pacific Heights have multiple layers of protection.

People who don't have security options need these tools most. Those who can afford gated communities with security guards would choose them if offered, but many can't afford such arrangements. The technology allows communities to make their own decisions about safety versus privacy tradeoffs.

A national Flock database or government facial recognition would be unacceptable, but community license plate readers with 2-week to 30-day retention periods are reasonable. An Illinois elected official who was initially anti-Flock became an advocate after becoming a victim of political violence and using Flock cameras to find the perpetrator.

The biggest damage from the PR disaster has been internal morale, with employees questioning why people are mad about things Flock doesn't do after nine years of operation. The business remains in good shape despite the challenges.

Flock raised additional funding at the beginning of the year. The company continues growing, though growth could accelerate with lower customer churn, which currently stands at 1%. The CEO emphasized the desire to return to zero churn through improved communication, better listening, and getting the right message to customers. At the core, communities want safety, which remains a fundamental human need alongside food and shelter.

Safety ranks high in Maslow's hierarchy of needs. Companies must lead proactively rather than play defense like Facebook and Zuckerberg did with Instagram's impact on teens, eating disorders, and social media addiction. Similarly, X and Twitter faced issues with political interference and attempts to subvert democracy. The CEO discussed Zuckerberg's recent manifesto addressing AI job displacement, advocating for open source tools and free AI access to level the playing field between rich and poor. CEOs must anticipate viral stories in the current media environment and own potential downsides proactively.

The CEO addressed Tucker's criticism and confirmed Flock appeared on the Today Show and Good Morning America that morning. The company seeks meaningful conversations with communities rather than specific podcast appearances. The business includes multiple product lines beyond license plate recognition, which represents just under 50% of forward-looking revenue.

Drones represent the fastest growing business unit. These systems deploy from HVAC control boxes on roofs, typically at police precincts or fire departments. When 911 calls occur, drones can be dispatched instead of officers first. In a Dallas-Fort Worth case, a woman reported a man walking with a gun at a gas station late at night. Rather than sending an officer with lights and sirens, a drone launched and arrived within a minute at 400 feet altitude traveling 60 miles per hour. Using 40x optical zoom, operators determined the object was a lighter, not a weapon, and reassured the caller. This demonstrates how drones mitigate potentially dangerous situations while enhancing community safety.

The company introduced an audit trail feature for drone operations. Third-party auditors could review drone flights to prevent misuse, such as stalking or privacy violations. Currently, city managers can access audit logs showing drone flights and patterns. Most customers have transparency portals displaying camera counts, drone flights, and audit logs, though active criminal investigations include a delay before public release. 911 calls are public record in many jurisdictions, and drone flights could potentially be published similarly.

New Hampshire maintains a three-minute retention period for license plate reading data. New York allows 21 days. Washington state's requirements were also referenced. The CEO noted minimal state regulation and zero federal regulation beyond air safety standards. The industry will need to establish default transparency standards proactively. Spectator mode exists for city officials and command staff to monitor operations.

Communities should decide whether to implement these systems, with options to disable entirely or set retention periods like 48 hours. The CEO emphasized that communities make these choices rather than companies imposing solutions.

Keep All-In Podcast in your library

Save the videos and channels worth coming back to, and find them again in one place.