AI/ML●●Solid
Jacobian Fingerprinting in LLM's
Jacobian readouts reveal coder identity inside model layers, not just outputs.
Big BrainNiche Gem
tristenharr
2011d ago
Back to browse
Whorl – Fingerprinting LLMs as horrible password generators
by tehryanx·Mar 25, 2026·2 points·0 comments
AI Analysis
●●●●GemBig BrainZero to OneNiche Gem
Identifies LLM models by password bias patterns when they refuse to tell you.
Strengths
- •Novel fingerprinting technique bypasses model identity lies during red team assessments.
- •Achieves 92% identification accuracy with just five generated password samples.
- •Practical red teaming tool for opaque API deployments where models hide identity.
Weaknesses
- •Limited to text generation biases, may degrade as models improve randomness.
- •Requires maintaining pattern database as new models release.
Category
Target Audience
Security researchers, AI red teamers
Similar Projects
AI/ML●●Solid
Lmscan – Detect AI text and fingerprint which LLM wrote it (zero deps)
Model attribution beats GPTZero's binary AI-or-not verdicts.
Niche GemShip It
zach22
303mo ago
Security●Mid
Anonymize LLM traffic to dodge API fingerprinting and rate-limiting
OHTTP proxy for OpenClaw, but only tested on two providers and limited adoption.
Bold Bet
clawshield
104mo ago
Other●Mid
Tool that hashes files into poems
Fun novelty turning file hashes into poems, explicitly not for production.
Cozy
0x12A
111mo ago
AI/ML●Mid
I built a free LLMs.txt generator
Free llms.txt generator when several already exist for this emerging standard.
Ship It
aiwrita
522mo ago
AI/ML●●Solid
Pokémon SVG Generation LLM Benchmark
Finally, a benchmark that uses Pokémon to test if models understand complex geometry.
Niche GemCrowd Pleaser
haxfenx
202mo ago