Distinguish steganography, watermarking, authorship analysis, and statistical evidence while testing benign fixed-marker carriers and transformation resilience.
Linguistic steganalysis looks for evidence that ordinary-looking prose carries a second machine-readable channel, while text watermark detection looks for an intentionally embedded provenance signal; neither detector result provides absolute certainty.
The submitted report remains byte-identical. This separate review, checked 2026-08-26, narrows implementation claims and gives visitors a decision path before they generalize from a local result.
Output produced deterministically by the local runtime.
Useful model or subset that must not be mistaken for full conformance.
Conclusion that needs an exact implementation, trust system, model, parser, or human review.
| Question | What the local page can answer | What it does not establish | Next evidence step |
|---|---|---|---|
| Does the fixed benign fixture contain the allowlisted marker pattern? | Yes — the deterministic view can answer that narrow question. | A general covert channel or secret payload. | Use the exact published method and controlled corpus if authorized. |
| Is unusual style evidence of steganography? | It is only a review indicator. | Intent, key use, or payload extraction. | Compare matched controls and transformations. |
| Is text watermarked? | Not by this laboratory. | Any provider watermark or provenance claim. | Use the specified detector, key, tokenizer, and calibrated threshold. |
Contrasts a known prepared pattern with naturally occurring indicator noise.
Expected boundary: Only the fixed marker fixture produces the allowlisted marker finding; both retain uncertainty language.
Open prepared laboratoryNo single view is automatically authoritative. Preserve the source, identify each parser or receiver, and compare their outputs before authorizing a consequential decision.
The ordinary surface prose presented to a reader.
Acrostics, positions, word lengths, punctuation, or lexical classes.
Distribution shifts, perplexity, token partitions, and detector scores.
A keyed or model-specific provenance indicator.
Edits, paraphrases, translation, normalization, and retokenization.
What the observed indicator supports and what remains unproven.
The workflow preserves evidence before transformation and keeps structural inspection separate from execution, remote verification, or model behavior.
These controls are contextual. They reduce a defined risk; they do not guarantee safety, truth, attribution, or resistance to every adaptive attack.
Use multiple independent indicators rather than one opaque score.
Keep provenance assertions separate from authorship and truth claims.
Test transformations representative of the real transport path.
Avoid publishing unrestricted arbitrary-payload encoders.
Retain human review for consequential attribution decisions.
Definitions are linked into the site-wide glossary and back to the full report.
The amount of hidden information carried per word, token, or carrier choice.
The ability of a channel or watermark to survive specified transformations.
The degree to which a detector can distinguish carrier text from an expected distribution.
The ordinary-looking text that hosts a hidden channel.
The proportion of benign examples incorrectly classified as containing a target signal.
An edit such as paraphrase, translation, normalization, or retokenization intended to disrupt a signal.
A method that tests for a specific embedded provenance signal.
A difference in useful information available to a keyed receiver compared with an unkeyed observer.
These links are provided for visitors who want the governing specification, paper, framework, or implementation documentation. Links open in a new tab; the site does not fetch them during runtime analysis.
Probability-aware generative linguistic steganography.
Original research or formal conference publication.In-context and zero-shot generative channel research.
Original research or formal conference publication.Tokenization drift in steganography and watermarking.
Original research or formal conference publication.Watermarking, detection, and provenance background.
First-party guidance, framework, registry, or standards-program material.Evaluation resources for media forensics.
First-party guidance, framework, registry, or standards-program material.Black-box and prompt-based generative steganography research.
Prepublication research that should be checked for later peer-reviewed versions.Survey of intentions, techniques, and challenges.
Useful supporting material that does not replace the governing specification or original study.The full submitted report is preserved byte-for-byte in the governed research library and in durable repository documentation. The laboratory turns selected concepts into deterministic local output without external calls or hidden persistence.
A defensive analysis of information-theoretic security, linguistic carriers, generative steganography, text watermarking, detector uncertainty, transformation channels, tokenization inconsistency, and safe enterprise review.
Read governed reportInspect prepared benign text for acrostic, fixed-position, word-length, punctuation, lexical, and transformation indicators using only the fixed marker MACHINE_TRADECRAFT_DEMO_OK.
Open bounded laboratory