Three researchers from Columbia University have created a technique named FontCode that can be used to embed hidden messages inside font glyphs (characters).

The technique takes advantage of how computers work with font glyphs, which for them, are nothing more than mathematical equations used to draw lines and curves on a screen.

FontCode introduces perturbations in normal text graphics

The FontCode technique consists of altering these equations to produce slight perturbations in one or more font characters.

An external observer can scan for these perturbations and decode them into indivudual letters based on a custom algorithm.

FontCode scheme

According to researchers, FontCode perturbations were designed to produce very slight alterations to current font glyphs, so they remain imperceptible to the human eye of a third-party observer. The entire idea is to hide and relay hidden messages inside normally-looking text.

Researchers say the FontCode algorithm could be used to hide secret messages inside all kinds of text, not just alpha-numeric characters. For example, music notes or mathematical equations printed inside research papers.

FontCode is also a medium-agnostic technique, because font perturbations could be embedded and later read from text that has been deployed in various mediums, such as images, PDF documents, Word files, screens, printed paper, and so on.

FontCode can replace QR code and watermarking systems

The team hopes their FontCode system would be adopted in the real world. They say the system could replace barcodes and QR codes and could be used to embed additional information inside labels, where there's no additional space for a big barcode or QR code.

Second, the system could also be used by whistleblowers and leakers to embed hidden messages inside benign-looking text that would not arouse suspicion during a visual inspection.

Furthermore, the system could also be used as an anti-whistleblower, document watermarking, or as a copyright infringement detection system by embedding hidden messages in text or works of art that might get leaked, allowing data owners to track down the origin of the leak based on the hidden message embedded in the leaked document.

For now, the system is still under development, but researchers plan to release tools for embedding hidden messages in font glyphs, and a mobile app to scan text, look for perturbations, and decode the hidden text. A demo of this app and an explanation for the FontCode technique is embedded below.

More details are available in a research paper titled "FontCode: Embedding Information in Text Documents using Glyph Perturbation" [1, 2].

Related Articles:

ZeroFont Technique Lets Phishing Emails Bypass Office 365 Security Filters

Mobile Devs Making the Same Security Mistakes Web Devs Made in the Early 2000s

Nvidia Creates AI for Training Robots to Learn From Watching Humans

Voice Squatting Attacks Impact Amazon Alexa and Google Home Assistants

Smarter People Don’t Have Better Passwords, Study Finds