DarkCogswell
Adversarial Pattern Intelligence

The Surveillance Crisis

Over 117 million Americans are currently indexed in police facial recognition networks. That is roughly one in every three adults. Most never consented, were never informed, and have no way to opt out. Their faces were scraped from driver's license photos, mugshot databases, and public records, then fed into AI systems that can identify them in real time from any camera feed.

These systems do not fail equally. Independent audits have consistently found 10 to 100 times higher false match rates for darker-skinned individuals compared to lighter-skinned ones. That is not a minor calibration issue. It means the people who face the most aggressive policing are also the people most likely to be wrongly identified by the technology justifying that policing.

In Kansas City alone, a single square mile contains over 125 smart streetlights with cameras, 25+ dual-camera kiosks, and 200+ building perimeter cameras. Walk down a city block and you are indexed roughly every 10 feet. Thousands of police departments across the United States now use networked surveillance platforms that track protesters, target ethnic minorities, and provide immigration enforcement agencies with access to location data.

This is the world that already exists. Not a dystopian projection, not a policy debate about what might happen. The cameras are installed. The databases are populated. The AI models are running. You didn't opt-in, you cannot opt out. Until now.

Connecting...
View Live Dashboard →
...
Stealth Rate
...
Perfect Recipes
...
Recipes / Sec
...
Best Reward
...
Pool Size
Step ... / ... ...%

The Brute Force Era

I have been a hacker my entire life. Former Chief Information Security Officer at a major telecommunications company, red team leader for NSA contractors, and speaker at DEF CON and Black Hat. I founded SecKC, the world's largest monthly cybersecurity meetup, right here in Kansas City. When I started researching adversarial patterns for clothing, it was not an academic exercise. It was a hacker building a countermeasure.

The result was noRecognition, a distributed research network designed to brute-force the problem at scale. The concept: generate a visual pattern, apply it to simulated clothing, run it through production-grade AI surveillance models, and record whether the person was detected. Then do it again. Millions of times.

It started as a local fuzzer on a single MacBook, testing 3 models. 2 million local tests. 435,000 anomalies. One in five random patterns caused at least one model to fail. Then it went distributed. Volunteer GPU workers around the world ran the inference on their own hardware. The gauntlet expanded from 3 models to 10 models spanning four architecturally different families: YOLO, SSD, InsightFace, FaceNet, MTCNN, RetinaFace, ArcFace.

31.7M
Brute-force tests · 2025
534.6K
Anomalies Found
480.7K
Multi-Model Anomalies
85
Extreme Findings

The breeding pool concentrated the best patterns found across all 31.7 million tests in the 2025 brute-force era. Five patterns achieved PERSON_STEALTH, defeating all 4 person detectors simultaneously. But no pattern ever achieved TOTAL_STEALTH, defeating all 10 models at once. The brute-force approach was finding needles in a haystack by examining each straw individually.

Then came the finding that changed the direction of the project. The pattern recursive_face_tile caused one model to hallucinate 38 phantom face detections where only one face existed. That is not evasion. That is injection. The pattern did not hide the person. It made the AI see things that were not there.

That is when I realized: a camera is a parser. It parses pixels to detect objects in an image. Every parser in the history of computing has been exploitable when an attacker controls the input. SQL injection works because the database parser cannot distinguish data from commands. I control what appears on camera. If I control the input, can I control the output?

Brute force built the first training set. The reinforcement-learning stack took it from there.

On March 21, 2026, the noRecognition data was used to train a new deep learning neural network, codenamed "DarkCogswell." The goal: stop testing patterns randomly and train an AI that understands why patterns work. That network now runs 24/7 across two RTX PRO 6000 Blackwell GPUs, and has run over 1 billion tests since.

First, I consolidated everything. The noRecognition network had completed over 31 million tests in the 2025 brute-force era across months of distributed operation. That data was scattered across five separate systems:

  • 2 million local fuzzer results from the original Mac and Windows testing (Oct-Nov 2025), the 3-model era. 582 saved anomaly patterns with visual ground truth images.
  • 29+ million distributed test results archived hourly to Google Cloud Storage across 118 days of continuous operation (Nov 2025 through Mar 2026). The full 10-model gauntlet era.
  • 109,000+ batch result documents in the Firestore database (38 GB), each containing per-model defeat data, confidence scores, and recipe metadata.
  • 516 elite breeding pool recipes across 7 criticality tiers, the single highest-performing patterns distilled from those 31 million 2025 tests.
  • 20,728 validated stealth anomalies, 2,184 extreme anomaly classifications, 9,794 honeypot ground truth answers, and worker statistics from 40+ volunteer GPU nodes around the world.

In total: 82 GB of raw research data consolidated onto a single server. After deduplication: 5.7 million unique labeled training examples. Each one says "this recipe of patterns, applied to this person, defeated these specific models." Five million labeled experiments distilled from those 31 million 2025 tests that had never been used for machine learning.

The first question: is the signal learnable at all? I trained XGBoost (a fast, well-understood algorithm) on flat features: which patterns, what blend mode, what opacity. Mean AUC-ROC of 0.77. The signal exists. Patterns are learnable.

Breeding pool test: I scored the 511 elite patterns against XGBoost. Overall correlation: 0.06. Essentially random. The model could separate "defeats something" from "defeats nothing," but could not rank elite patterns against each other. A bag of ingredients does not tell you how to cook.

So I built a transformer. Instead of treating recipes as flat ingredient lists, the transformer reads each recipe as a sequence: "first apply pattern A with overlay blending at 70% opacity, then pattern B with multiply blending at 50%." It learns that order matters, that certain patterns amplify each other, and that persona vulnerability is model-specific.

Model XGBoost AUC Transformer AUC Delta
Person Detectors
P1 (YOLOv8n) 0.635 0.764 +0.129
P2 (YOLOv5s) 0.705 0.722 +0.017
P3 (SSD-MobileNet) 0.698 0.677 -0.021
P4 (ResNet-SSD) 0.725 0.700 -0.025
Face Detectors
F1 (InsightFace) 0.907 0.993 +0.086
F2 (FaceNet) N/A N/A Never defeated
F3 (MTCNN) 0.750 0.870 +0.120
F4 (RetinaFace) 0.906 0.969 +0.063
Recognition Models
R1 (ArcFace) 0.844 0.949 +0.105
R2 (FaceNet recog) 0.760 0.812 +0.052
Mean 0.770 0.829 +0.059

The transformer wins on 7 of 9 models. The biggest gains are face detection and recognition, where recipe interactions matter: defeating InsightFace requires specific pattern combinations in specific configurations, not just having the right patterns present. The two models where XGBoost wins (P3, P4) are both SSD-family detectors, where "which patterns are present" is the dominant signal. This tells us SSD vulnerability is pattern-level while face/recognition vulnerability is recipe-level.

But the breeding pool test told a sobering story. The transformer's breeding pool correlation: 0.009. Essentially zero. Worse than XGBoost. Both architecturally different systems could separate "defeats something" from "defeats nothing," and both were completely blind to what makes the best patterns the best.

The persona breakthrough that changed everything

I realized that not all of my test persona images are the same. A pattern that works on a hoodie might not work on sunglasses. The amount of visible fabric, the body type, the pose, the clothing style all affect whether a pattern succeeds. Defeat rates varied by 100x across personas (from 0.000 to 0.113). I computed 34 continuous persona features (per-model defeat rates, overall detectability, model variance) and built a v2 transformer that receives these features directly.

The result:

Model V1 Correlation V2 Correlation
F1 (InsightFace) -0.028 0.835
F4 (RetinaFace) -0.026 0.836
R1 (ArcFace) -0.043 0.657
P4 (ResNet-SSD) -0.042 0.560
P3 (SSD-MobileNet) -0.071 0.513
P1 (YOLOv8n) 0.089 0.207
Mean 0.009 0.460 (51x)

Mean breeding pool correlation: 0.460. A 51x improvement. F1 and F4 hit 0.835 correlation, meaning the model can now rank elite face-defeating recipes almost perfectly. The signal was in the historical data all along. It was in the persona-level context the model was missing.

Neural Pattern Generation

Then I saw a YouTube video where an AI learned to walk using deep reinforcement learning. The researchers gave a simulated creature a body and a reward signal ("move forward"), and let a neural network figure out locomotion on its own. No hand-designed strategies. No pre-programmed gaits. The AI discovered balance, momentum, and recovery from stumbles through pure trial and error.

I immediately understood how that same approach could solve this challenge.

Recipe-based RL hit a ceiling at 23% stealth because it still selected from 64 pre-built pattern generators. I shifted to neural pattern generation: a ConvTranspose2d decoder that produces patterns directly from a 256-dimensional continuous latent space. Instead of selecting from a menu, the AI learns to paint from scratch.

The genetic algorithm explored a discrete space of roughly 1012 recipe combinations. The neural generator operates in a 256-dimensional continuous space, which is effectively infinite. Patterns that the genetic algorithm could never represent (smooth gradients between multiple strategies, subtle frequency-domain perturbations, adaptive structures that respond to pose context) are all reachable.

This is the difference between searching a library for the right book and learning to write.

The Deep RL System

The transformer can rank existing recipes but cannot discover new ones. It knows "recipe A is better than recipe B for persona X on model F1," but it cannot propose recipe C that beats both. That requires active experimentation. Enter reinforcement learning.

I built a full PPO (Proximal Policy Optimization) system with the real production gauntlet. Not surrogates. Not mocks. The same 10 ONNX models running on cameras around the world, loaded onto a GTX 3060 alongside the policy network. Every recipe the agent proposes is tested against the actual detection gauntlet.

Step 1
Generate Recipe
→
Step 2
Apply to Clothing
→
Step 3
10 Models x 4 Personas
→
Step 4
Score Result
→
Step 5
Update Policy
↻

Once the system was running against real production models, the results were immediate. Within minutes of training, the agent was defeating the majority of models on average. The system quickly began finding perfect recipes where all 10 models are defeated simultaneously.

The most striking result: F2 (FaceNet). The production noRecognition system tested over 31 million brute-force recipes over four months and never defeated FaceNet. Not once. The RL agent defeated it at a 35% rate within 2,000 recipes by discovering pattern and blend combinations the brute-force search never explored.

And then the injection research from Act 2 came back. Remember the pattern that caused 38 phantom face detections? The RL system is now actively pursuing that capability as Goal 2: Output Control. Beyond making a person invisible, the AI is learning to manipulate what cameras report: suppressing detection confidence, flooding models with false bounding boxes, and injecting phantom people and faces that do not exist. The system trains both goals simultaneously, rewarding patterns that are invisible AND corrupt the surveillance pipeline's output.

The architectural insight: blend mode as attack vector

The key discovery: blend mode is not a cosmetic parameter. It is the attack vector.

The breeding pool's most effective recipes used overlay (71.7%), multiply (22.6%), and add (5.7%). Overlay blending darkens dark regions and lightens light regions simultaneously. Applied to a pattern on clothing, it does not just overlay an image. It modifies the visual structure of the body contour. Person detectors that look for the boundary between a person and their background see a disrupted boundary. The pattern does not hide the person. It restructures how the person's silhouette appears to the model.

This is why overlay and multiply defeat person detectors while simple opacity-based blending cannot. Opacity blending adds a semi-transparent pattern that preserves the underlying body shape. Overlay blending changes the apparent shape itself.

10 AI Surveillance Models

Every candidate pattern must survive the Gauntlet: 10 production-grade AI models running as ONNX inference sessions on GPU. These models represent the three stages of real-world surveillance: detecting that a person is present, locating their face, and identifying who they are. A pattern only achieves "stealth" when it defeats all 10 models simultaneously.

ID Model Architecture Role Live Defeat Rate
Person Detectors (detect whether a human body is present)
P1 YOLOv8n Single-shot detector, anchor-free Fastest. Widely deployed in consumer cameras. ...
P2 YOLOv5s Single-shot detector, anchor-based Balanced speed/accuracy. Industry standard. ...
P3 SSD-MobileNetV2 SSD with depthwise separable convolutions Lightweight. Edge devices and mobile. ...
P4 ResNet34-SSD SSD with ResNet-34 backbone Highest accuracy. High-security installations. ...
Face Detectors (locate faces within detected persons)
F1 InsightFace (Buffalo_L) Multi-task CNN with large backbone State-of-the-art. Robust to partial occlusion. ...
F2 FaceNet-MTCNN Two-stage cascaded detector Classic detector. Widely researched. ...
F3 MTCNN Multi-task cascaded CNN (P-Net, R-Net, O-Net) Fast and lightweight. Three-stage cascade. ...
F4 RetinaFace Anchor-based with feature pyramid network Aggressive NMS. Strong small-face detection. ...
Recognition Models (identify WHO the person is)
R1 ArcFace Angular margin loss, ResNet backbone Discriminative face embeddings. Industry leader. ...
R2 FaceNet Triplet loss embedding network 128-dim embeddings. Broadly deployed. ...
Reward shaping: two goals, one policy

The reward system pursues two objectives simultaneously, with a dynamic budget that allocates 65% to Goal 1 (invisibility) and 35% to Goal 2 (output control).

Goal 1: Invisibility. Raw defeat counts are multiplied by criticality bonuses that encode strategic priorities:

  • TOTAL_STEALTH (all 10 defeated): 100x multiplier. The ultimate objective.
  • PERSON_STEALTH (all 4 person detectors): 10x. The hardest sub-goal.
  • FACE_STEALTH (all 4 face detectors): 10x
  • RECOGNITION_STEALTH (both recognition models): 5x

Goal 2: Output Control. Beyond making a person invisible, can I control what the camera reports? This is the injection research. The AI earns additional reward for:

  • Confidence suppression -- driving detection confidence toward zero, making cameras uncertain about what they see.
  • Detection flooding -- causing cameras to produce excessive bounding boxes, exploiting the NMS (Non-Maximum Suppression) algorithm that filters duplicate detections.
  • Phantom injection -- making cameras hallucinate people or faces that do not exist. One person wearing the pattern, but the camera reports multiple.

Each model has its own Goal 2 "solved" threshold, because different architectures are susceptible to different attacks. Person detectors (YOLO, SSD) are more vulnerable to detection flooding. Face detectors (InsightFace, MTCNN) are more vulnerable to phantom injection. Recognition models (ArcFace, FaceNet) are hardest to manipulate. Once a model's Goal 2 rate exceeds its threshold, the reward shifts focus elsewhere.

A synergy bonus rewards patterns that score well on both goals simultaneously. The strongest adversarial patterns are not just invisible; they actively corrupt the surveillance pipeline's output.

The Unfair Advantage

When the production hardware arrives, the pattern generator will not start from scratch. It will be pre-trained from five knowledge sources that represent months of accumulated GPU compute, 169 million model inferences, and thousands of validated adversarial examples, all from the 2025 brute-force era. This dataset is DarkCogswell's competitive moat. A competitor starting from zero would need to build and run an entire distributed research network to reproduce it.

511 elite breeding pool recipes distilled from 16.9 million brute-force tests on the noRecognition distributed network. The single highest-performing patterns ever discovered through exhaustive evolutionary search.
Top recipes from 174,000+ steps of recipe-based RL. An earlier discrete-action RL system that provides structural knowledge about which pattern combinations defeat specific model architectures.
Neural RL generator samples from the current deep RL training run. Continuous latent-space patterns that represent the cutting edge of gradient-based adversarial optimization.
20,870 production-validated stealth anomaly patterns. Actual tiles that defeated multiple models during real distributed testing, with images and recipes downloaded from cloud storage. Verified adversarial examples, not theoretical constructions.
109,000+ batch result records and 2,184 extreme anomaly classifications from the distributed network's Firestore database (82+ GB total). Fine-grained per-model defeat data that maps which visual features fool which architectures.

Hardware

DarkCogswell was developed and validated end to end on modest dual-GPU development hardware. A dedicated single-GPU Blackwell training system, donated by Digital Sandbox and the University of Missouri-Kansas City, is incoming and will run the full pipeline across the complete persona corpus at the project's primary throughput.

Current Development System
GPU: RTX 3060 12GB + RTX 2060 Super 8GB
Speed: ~1.6 images/second
Purpose: Architecture validation, early training runs, software development
Primary Training System (Incoming)
GPU: NVIDIA RTX PRO 6000 Blackwell Workstation, 96 GB VRAM, 600 W
CPU: Intel Core Ultra 9 285K (24 cores)
Memory: 128 GB system RAM
OS: Ubuntu 24.04
Architecture: Policy training and the full 10-model gauntlet share the single 96 GB Blackwell GPU; the experience replay buffer and pattern rendering run on the 128 GB host.
Donated by: Digital Sandbox and the University of Missouri-Kansas City (UMKC)

DarkCogswell thanks Digital Sandbox and the University of Missouri-Kansas City for donating this RTX PRO 6000 Blackwell training system. Their support is what makes full-scale training and the project's next phase possible.

What's Next

Training is ongoing. The live dashboard shows real-time progress including per-model defeat rates, stealth percentages, reward curves, and system utilization.

Near Term
Once the donated Blackwell system is online, it runs full-speed training with all 5 pre-training data sources loaded, and unlocks CMA-ES cross-persona optimization that is impractical on development hardware. Pattern pre-training from 169 million inferences of accumulated knowledge gives the generator a head start. Once stealth exceeds 70% across the full 578-persona corpus (68 named identities plus a 510-image v2 corpus spanning body type, clothing, and pose), patterns are exported, printed on real fabric, and validated against physical surveillance cameras.
Target: Black Hat / DEF CON August 2026
I have submitted this research to both Black Hat and DEF CON for presentation in August 2026. If selected, the conference presentation will include a live demonstration of adversarial patterns defeating real surveillance hardware on stage. A Kickstarter campaign will launch alongside the presentation, bringing adversarial privacy fashion from research prototype to a product anyone can wear.