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

Episode Overview

• This episode explores the critical tension between public safety and individual privacy through the lens of Flock Safety, a leading provider of License Plate Reading (LPR) technology.
• It examines how modern, objective vehicle-tracking technology can solve the vast majority of crimes within a short window while deliberately avoiding invasive biometric tracking like facial recognition.
• The narrative traces how technology providers can design ethical guardrails, including automated audit systems, short default data retention periods, and zero-trust principles, to prevent insider abuse by law enforcement.
• It highlights the importance of local democratic oversight, arguing that elected officials—not tech vendors—should ultimately determine the balance of security and privacy within their own communities.

Key Concepts

• License Plate Reading (LPR) Technology: Flock Safety's core technology relies on capturing still images of vehicles, reading license plates, and identifying unique vehicle characteristics (e.g., roof racks, bumper stickers, dents) rather than recording continuous video, looking inside vehicles, or using facial recognition.
• The Privacy vs. Safety Trade-Off: Implementing public safety technology requires balancing community security with individual privacy concerns. Determining the optimal duration for data retention (e.g., 7 days vs. 30 days) is a critical lever in balancing these interests, as longer retention increases privacy risks with diminishing returns for safety.
• Democratic Governance of Tech: Decisions regarding the deployment, parameters, and data retention policies of public surveillance tools are most appropriately made by elected local officials (e.g., city councils and mayors) who are directly accountable to their communities, rather than by private technology vendors.
• Mitigating Insider Abuse: Any system of power or surveillance is susceptible to abuse (e.g., officers stalking individuals). Protecting against this requires automated auditing tools ("Audit Assistance") using machine learning to flag anomalous search patterns rather than relying solely on easily ignored manual audits.
• Surveillance Data as a Liability: Stored surveillance data should be viewed by operators and technology companies as a legal and ethical liability rather than an asset. Keeping data longer than necessary increases the potential for abuse, retroactive surveillance, and public distrust.
• The Chilling Effect of Predictive Surveillance: Utilizing AI to identify "abnormal" behavior or predict crimes before they happen presents a severe threat to civil liberties. A strict human-in-the-loop framework must be maintained, and automated systems should never be allowed to dynamically define suspicion.

Quotes

• At 2:49 - "A car drives by, we take a still image of the vehicle, we read the license plate... the system's gotten more sophisticated with time, we can detect things like a roof rack, or a bumper sticker, or a dent, or damage." - Explaining how Flock's technology relies on objective vehicle metadata rather than invasive biometrics.
• At 4:15 - "I want to live in [a society] where we're safer, and I want to keep my privacy while I do it, but that is hard... to find that balance is not easy." - Highlighting the core tension between civil liberties and public safety.
• At 7:21 - "This data is a liability, not an asset. You should have a set retention period." - Advocating for strict data deletion policies to limit the potential for retroactive surveillance and data leaks.
• At 8:42 - "We think 90% of crimes will be solved within seven days of data retention." - Providing empirical context on how shorter retention windows can still preserve the vast majority of the technology's investigative utility.
• At 10:17 - "What's important... is making sure city councilors and mayors are educated that they have control. They can make this choice." - Emphasizing the necessity of local democratic oversight over technology deployment.
• At 12:20 - "We don't do facial recognition, we don't capture video, we don't look inside the car, we don't even allow in our system to search for people... and that's in direct conflict with some of our biggest competitors." - Differentiating Flock's privacy-by-design constraints from more invasive surveillance platforms.
• At 20:25 - "We also kind of know what abuse looks like... if you're stalking me... you would search for me again [day after day]. That's very abnormal. If you were truly trying to find me because I committed a crime, you would have put my tag on a hot list." - Describing the behavioral heuristics used to automatically detect unauthorized system use by police officers.
• At 33:51 - "We've been really consistent that we never want to be a company that defines suspicion. That's a really dangerous path." - Outlining the ethical boundary of refusing to let automated AI systems dynamically label citizens as "suspicious" without human oversight.
• At 35:50 - "If a customer doesn't want to be held accountable, I don't think I want them as a customer... I'd rather lose [them]... I want to sleep at night." - Demonstrating the corporate responsibility of technology vendors to fire clients (including police departments) who refuse to implement basic ethical and privacy-preserving guardrails.

Takeaways

• Implement Short Retention Windows: Approximately 90% of crimes utilizing license plate databases are solved within a 7-day window. Organizations should default to a 7-day retention period to dramatically reduce the global footprint of stored data while preserving its core investigative utility.
• Focus Surveillance on Objects, Not Humans: To respect civil liberties, design public safety systems to track vehicle metadata (license plates, make, color, damage) rather than human biometrics (facial recognition) or continuous video feeds.
• Deploy Automated Oversight: Because manual reviews of database audit logs are mathematically impossible due to sheer volume, organizations must integrate machine learning and automated behavioral heuristics to flag abnormal search patterns and catch insider bad actors.
• Enforce Corporate Ethical Boundaries: Technology vendors must be willing to walk away from and fire high-value clients (including law enforcement agencies) who refuse to adopt basic transparency, accountability, and ethical auditing standards.
• Empower Local Democratic Decision-Making: Avoid one-size-fits-all national policies for public safety technology. Instead, educate and empower local city councils and mayors to set the specific access and retention policies that reflect their community's unique risk tolerance and values.