This technique uses a custom OpenType substitution system to display text that is different from the actual text contained in the document. Instead of editing or replacing the underlying characters, the font itself intercepts specific phrases and substitutes them with new ones at render time. This is achieved through OpenType GSUB rules, which allow the font to rewrite glyph sequences whenever certain font features—such as contextual alternates (calt)—are enabled.
Because the substitutions are handled entirely by the font, the true text remains unchanged in the source. Applications, scripts, and assistive technologies still read the original words, while the user sees a completely different message on the screen. The person reading the content has no indication that the displayed text does not match the underlying data unless they inspect the raw HTML code or disable the font feature.
In effect, this creates a visual deception mechanism: the font can make the viewer believe they are seeing one statement while the actual stored text says something else. By toggling a single font feature, the display can shift between the genuine content and the manipulated version, allowing the system to intentionally mislead the user without altering the document itself.