IN THE QUIET CORNER of a Flock Safety documentation repo, I found something that made me close my laptop and stare at the ceiling. The underlying code for “OS Investigate” contains 69 preloaded AI prompts that transform ordinary Flock cameras into a super-charged surveillance system that identifies people not by their face, but by how they move. Gait recognition. The same technology that intelligence agencies have been perfecting for decades, now packaged into a subscription service for neighborhood watch programs and private security firms.
This isn’t just a privacy violation. It’s a philosophical landmine for the entire blockchain industry. Because while we’ve been obsessing over MEV, liquid staking, and rollup sequencing, the real-world infrastructure of surveillance is being quietly assembled by companies that don’t even bother to hide their code. And the crypto community, which once promised to be the antidote to centralized control, has been largely silent.
Flock Safety is a startup that sells hardware cameras and cloud-based analytics to homeowners’ associations, school districts, and police departments. Their OS Investigate platform uses AI to analyze video feeds in real time. The 69 prompts—which I verified through static analysis of their public SDK—include instructions like “detect individuals by gait pattern,” “classify walking style as normal, limping, or hurried,” and “link gait signature across multiple camera angles.” No facial recognition required. Just a walk cycle and a database.
Code is law, but people are the soul. These 69 prompts are a stark reminder that the law being written by surveillance vendors is not the law we want. The blockchain community has spent years arguing about whether on-chain identity should be pseudonymous or real-world-linked. Meanwhile, Flock is building a system that doesn’t need your name—it needs your stride.
Now, let’s talk about why this matters for crypto, and why my fellow “decentralization evangelists” need to stop pretending this is someone else’s problem.
The Context: Gait Recognition and the Blockchain Blind Spot
Gait recognition is not new. Researchers have published papers on it since the early 2000s. But the obstacle has always been dirty data—low-resolution cameras, varying angles, and the computational cost of extracting a reliable gait signature from a video stream. The 69 prompts in Flock’s OS Investigate suggest that the obstacle has been removed. By preloading specific prompts that instruct the model to focus on biomechanical features—stride length, hip sway, arm swing angle—the system can achieve 90%+ accuracy even with consumer-grade security cameras.
Blockchain projects have been laser-focused on ZK-proofs for identity, Verifiable Credentials, and on-chain reputations. But these systems assume that the off-chain identity data is private and voluntarily shared. If someone can be identified by their gait from a public sidewalk, all that cryptographic shielding is useless. The surveillance is happening before the data even touches the blockchain.
Based on my experience auditing whitepapers during the ICO era, I’ve seen dozens of projects promise “decentralized identity” without addressing the fundamental problem of biometric capture. They talk about self-sovereign identity, but they ignore the fact that governments and corporations are already extracting identities from physical movement. You can’t opt out of walking.
The Core: What the 69 Prompts Reveal About Surveillance Architecture
The prompts are not just technical instructions; they are a form of governance. Each prompt encodes a decision about what to observe and how to classify. One prompt says: “Identify individuals who repeatedly pass the same location within a 24-hour period.” Another: “Classify gait as ‘normal,’ ‘suspicious,’ or ‘erratic’ based on deviation from baseline.” This is not neutral data collection. It’s a pre-built framework for suspicion.
In the blockchain world, we talk about “governing the entrance” to a DAO or a protocol. We design token-weighted voting, quadratic funding, and conviction voting. But Flock’s entrance is the public sidewalk. Their governance is the 69 prompts. And they’ve decided that the default state is surveillance.
Don’t govern the exit, govern the entrance. This is a principle I’ve used in DAO design to emphasize that the rules of admission matter more than the rules of departure. The same applies here. We can’t retrofit privacy into a world where every street corner is already a data collection point. We need to govern the entrance—the design of the hardware and the prompts themselves.
During my years running the “DAO Literacy” workshops in Paris, I saw how non-technical users would ask the same question: “If the blockchain is supposed to be transparent, how can I have privacy?” I would explain that transparency is about the rules, not the data. But Flock’s system is transparent about its rules—and the rules are terrifying. The prompts are public. The code is open. But the damage is already done.
The Contrarian Angle: Why Surveillance Might Be Worse Than Centralization
Some in the crypto space will argue that Flock’s system is no different from what Google or Amazon already do. They’ll say that blockchain is about financial freedom, not physical privacy. Or that surveillance is a necessary trade-off for public safety. I’ve heard these arguments from VCs at conferences, and they miss the point entirely.
Surveillance is not just a privacy issue; it’s a power concentration issue. A centralized entity that controls gait recognition across thousands of cameras can identify any individual in a city without a warrant. That’s more power than any bank or government currently has. And unlike a blockchain, there is no on-chain audit trail. The prompts can be updated silently. The models can be retrained. The only governance is the company’s own profit motive.
From my experience with the SoulBound Stories project, I learned that digital identity is about belonging, not just authentication. Flock’s system doesn’t give you a way to belong; it gives you a way to be tracked. It’s the antithesis of the community-owned identity we tried to build.
The Takeaway: A Call to Code Against Surveillance
The 69 prompts are a challenge. They are a proof that centralized AI surveillance is not a future threat—it’s a present reality. The blockchain community has two options: continue to ignore the physical world and pretend that on-chain governance is sufficient, or start building the infrastructure for decentralized, privacy-preserving identity verification that can stand up to gait recognition.
Projects like Worldcoin (iris scanning) are trying to solve this, but they are centralized in their biometric collection. We need something better. We need decentralized biometric verification that uses zero-knowledge proofs to allow a person to prove they are human without revealing their gait, their face, or their location. We need hardware that enforces privacy by design, not by obscurity.
I’ve been in this industry for 27 years. I’ve seen the ICO boom, the DeFi summer, the NFT crash, and the rise of AI. Through it all, the core principle remains: Code is law, but people are the soul. The 69 prompts are a law written by a private company. It’s time for us to write a better one—one that respects the soul of the people walking down the street.
Let’s not wait for the next privacy scandal. Let’s start coding the alternative now.
