Hide and Seek in Embedding Space: Geometry-based Steganography and Detection in Large Language Models
Hide and Seek in Embedding Space: Geometry-based Steganography and Detection in Large Language Models explores Develop a security tool leveraging mechanistic interpretability to detect steganographic threats in fine-tuned language models.. Commercial viability score: 5/10 in AI Security.
BUILDER'S SANDBOX
Build This Paper
Use an AI coding agent to implement this research.
OpenAI CodexAI Agent
Lightweight coding agent in your terminal.
Claude CodeAI Agent
Agentic coding tool for terminal workflows.
AntiGravity IDEScaffolding
AI agent mindset installer and workflow scaffolder.
CursorIDE
AI-first code editor built on VS Code.
VS CodeIDE
Free, open-source editor by Microsoft.
Recommended Stack
Startup Essentials
Estimated $10K - $14K over 6-10 weeks.
See exactly what it costs to build this -- with 3 comparable funded startups.
7-day free trial. Cancel anytime.
Discover the researchers behind this paper and find similar experts.
7-day free trial. Cancel anytime.
Related Papers
- Builds On ThisKolmogorov Complexity Bounds for LLM Steganography and a Perplexity-Based Detection ProxyViability: 3.0
- Builds On ThisA Decision-Theoretic Formalisation of Steganography With Applications to LLM MonitoringViability: 3.0
- Builds On ThisOn the Possible Detectability of Image-in-Image SteganographyViability: 4.0
- Builds On ThisSeamless Deception: Larger Language Models Are Better Knowledge ConcealersViability: 2.0
- Builds On ThisHiding in Plain Text: Detecting Concealed Jailbreaks via Activation Disentanglement
Related Resources
- Why is AI security important?(question)
- What is the focus of AI security research?(question)
- How do AI security measures protect systems?(question)
- AI Security – Use Cases(use_case)