How DocuSentinel Protects Your Data
A deep dive into the steganographic encoding, multi-layer detection mesh, and forensic audit systems that make DocuSentinel unique.
256-bit identity, woven into grammar
Each document receives a unique DSID composed of 8 structured fields: Tenant ID, Document ID, Classification Level, Author Hash, Timestamp, Version, Encoding Method, and Reed-Solomon ECC parity bits. This identifier is then embedded using three independent steganographic techniques.
Zero-Width Unicode Insertion
Four invisible Unicode characters (U+200B, U+200C, U+200D, U+2060) encode the DSID as base-4 digits at inter-word boundaries.
Statistical Whitespace Patterns
Subtle variations in spacing between sentences and paragraphs encode redundant DSID copies using statistical models.
Semantic Micro-Substitutions
Swapping synonyms ("however" ↔ "nevertheless") encodes bits that survive even AI-powered paraphrasing.
Every exfiltration path, covered
The detection mesh operates at three layers: OS-level clipboard monitoring via the endpoint agent, browser-level input field scanning via the extension, and network-level HTTPS inspection via the ICAP proxy. Detection latency averages 387ms at p95.
Cryptographic chain-of-custody
Every detection event is hash-linked to its predecessor using SHA-256, forming an immutable Merkle chain. Events are anchored to external timestamping authorities (RFC 3161) for non-repudiation. The result: a forensic record that holds up in court.
See it in the live demo