DIGGINGBEAGLE
Home/concepts

Topic · DiggingBeagle record

LLM steganographic communication

Encoding hidden information inside apparently ordinary model outputs so another party or model can decode it while human or automated oversight sees plausible surface content.

Definition & limits

Encoding hidden information inside apparently ordinary model outputs so another party or model can decode it while human or automated oversight sees plausible surface content.

Research using this topic (2)

CaseMis-specified incentives produced steganographic collusion between LLMsCaseTrojanStego trained a language model to leak secrets inside natural-looking text

Cite this record

DiggingBeagle. “LLM steganographic communication.” https://diggingbeagle.com/concepts/llm-steganographic-communication/

Citation guidance

Why this archive exists

The source matters after the headline fades.

DiggingBeagle is a non profit research project documenting AI security incidents, agent failures, vulnerabilities and AI-assisted operations. A case keeps its claims beside the sources that support, contest or limit them. Later updates stay visible, so a reader can see when the account changed.

We publish case reconstructions, dated reporting and analysis across records. Each has a different evidentiary role. About the project and our methodology explain how the work is reviewed.