Back to the portfolio

Intelligence operatives

CybersecuritySoftware & AIAerospace & DefensePrototype likely

This technology hides encrypted messages inside ordinary-looking text generated by an AI language model. A sender's original message is encrypted and then used to steer a language model to produce natural-sounding text in any target language, with no apparent connection to the original content. That cover text is split across multiple apps or communication channels to avoid detection. The receiver uses the same language model to reverse the process and recover the original message.

What you could build

A secure messaging platform or SDK for defense contractors, government agencies, and journalists that wraps existing chat apps and automatically converts sensitive messages into innocuous AI-generated conversation, making covert channels invisible to network monitors and censors.

Readiness: Prototype likely

Concept, described but not yet demonstrated. Lab validated, supported by experimental results in the patent. Prototype likely, the text describes a built, working embodiment.

Readiness is inferred from the patent text, not from a lab visit.

The record

Inventors
Kai Zeng, Massieh Kordi Boroujeny
Granted
August 4, 2026
Status
Granted patent
Patent number
12701140

Ready to talk?

the university handles licensing for this technology. Email us for an introduction patents@harbor.capital.

Sonar summaries are generated from public patent text and are not legal advice.