Intellectual Property

Why AI Keeps Making Everything Purple
The Origins and Amplification of Generative AI's Color Bias
On the Tailwind pipeline, the visited link sediment, the Helmholtz-Kohlrausch perceptual inflation of blue-violet, and the difference between a hex value and what it looks like on a screen. The easy-read version is visible. The full academic version is encoded inside the text.
read (html) read (pdf)
Tools
ZWC Decoder
Zero-width character steganography decoder. Paste text, extract hidden messages.
steganography unicode
Cryptography & Information Hiding
Black-box Collision Attacks on NeuralHash Perceptual Hash Function
Leblanc-Albarel & Preneel — COSIC, KU Leuven. IACR ePrint 2024/1869
perceptual hashing collision attacks neuralhash
Keyed-Verification Anonymous Credentials with Highly Efficient Partial Disclosure
Mirzamohammadi, Bobolz, Sedaghat, Heydari Beni, Abidin, Singelée, Preneel — COSIC. IACR Communications in Cryptology 2(3)
anonymous credentials selective disclosure kvac
IACR ePrint 2024/658
International Association for Cryptologic Research
cryptography
COSIC Research Overview
Computer Security and Industrial Cryptography — KU Leuven
research group ku leuven
Side-Channel Analysis & Machine Learning
P2W: From Power Traces to Weights Matrix — An Unconventional Transfer Learning Approach
Siyadatzadeh, Mehrafrooz, Mentens, Stefanov — arXiv, 26pp
side-channel power analysis transfer learning
Color Science & Perception
Helmholtz-Kohlrausch Effect — Bimodal Hue Dependency
Li, Xu & Luo (2026). Color Research & Application 51(2). DOI: 10.1002/col.70057
perception h-k effect color science
Modeling the HDR Display with XCR
Stolitzka, Agahian & Poynton (2025). Information Display 41(4). DOI: 10.1002/msid.1596
display technology hdr
Perceived Brightness of Chromatic Stimuli with Backgrounds of Varying Luminance
Avendano Martinez, Zuena & Pytlarz (2024). Color Research & Application 50(1). DOI: 10.1002/col.22949
brightness perception cam16
The Helmholtz-Kohlrausch Effect on Display-Based Light Colors
High, Green & Nussbaum (2023). Color Research & Application 48(2). DOI: 10.1002/col.22839
display colors h-k effect
Classical Ciphers
Caesar, Affine, Vigenère, Hill — Foundations
Parametric ciphers where the key is the payload. Mod-97 prime alphabet.
caesar affine vigenère hill
RSA Public Key Cryptosystem
Foundations of asymmetric cryptography
public key rsa
AI Training Data & Model Behavior
Distributional Convergence in Generative AI Outputs
Anthropic Applied AI team — documented in Claude Code frontend design skill
distributional convergence training bias
Model Collapse in Recursive Training
Shumailov et al. (2023–2024). Nature. Model autophagy disorder.
model collapse synthetic data
AI Color Palette Bias in Generated Posters
Rong (2024). Color Research & Application. 120 AI-generated vs 120 human-designed.
ai generation color bias
Steganographic Techniques
Zero-Width Unicode Steganography
4-character ZWC encoding: U+200B, U+200C, U+200D, U+FEFF → 2 bits per character
zwc unicode invisible encoding
CSS Hex-Color ASCII Encoding
Every hex color is also an ASCII string. Dual-valid values as art direction.
css hex encoding layer 3
Structural Encoding — Word Counts, Section Counts, Cross-Page Patterns
Document structure as covert channel. Plausibly deniable because every document has structure.
structural covert channel layer 7
Hash-Based Cross-Document Linking
SHA-256 prefix verification between pages. Tamper-evident, non-obvious, time-proof.
sha-256 verification layer 8
AI Archaeology
The Long Tail: Semantic Archaeology of 99 Single-Install Claude Code Plugins
Cooper-Rye & Claude, 2026. Three-register reading (logical, lateral, hallucinatory) of every name in the long tail.
semantic archaeology plugin analysis naming
The contents of this page are provided for reference. Some items are original work by Alexander Thomas Cooper-Rye. Some are published research by other authors, linked here for educational purposes. The distinction between these categories is left as an exercise for the reader.
atcooper.net — 2026