The new Ghost Font: the art of hidden messages invisible to artificial intelligence

An inventive designer named Eric Lu has introduced Ghost Font to the world — an innovative typeface that reimagines the traditional perception of text. This experimental font allows people to easily decode hidden messages while remaining elusive to modern artificial intelligence models. According to a specialized publication, the development is based on an understanding of human perception and the weak points of AI algorithms.

How it works: a play of light, shadow, and motion

Unlike classic fonts, where letters are formed by clear outlines, Ghost Font works according to a completely different aesthetic. Each text message is created using thousands of microscopic dots. The key lies in motion: the dots that make up the letters move synchronously in one direction, while the surrounding dots move chaotically or in the opposite direction. The human brain, naturally adapted to recognizing patterns in motion, intuitively groups these elements, making the hidden word easy to perceive. The visual effect is so subtle that when the animation stops, the text literally dissolves, leaving only a disorderly cluster of dots on the screen, devoid of any recognizable structure.

Why artificial intelligence is “blind to it”

The secret of Ghost Font’s resilience to AI recognition lies in one of the key characteristics of modern multimodal models. While the human eye is excellent at recognizing objects based on dynamic changes and movement, many advanced AI algorithms still analyze video content as a sequence of static frames. Since Ghost Font lacks clear, static outlines, which are the basis of traditional optical character recognition (OCR) systems, AI models often cannot correctly identify the hidden message. Their algorithms, tuned to analyze shape rather than motion, simply do not see text where a person can easily read it.

The project also makes it possible to add elements of “decoy text.” These are intentionally invented sequences that can mislead AI algorithms, forcing them to “see” words or phrases that do not exist, while the human eye continues to perceive the real, correct text. This feature highlights the differences between how humans and machines process information.

It is important to note that the creator of the experiment emphasizes that Ghost Font is not intended to be a reliable tool for encryption or protecting confidential data. Specialized computer vision algorithms that analyze motion in high detail, or even a large number of frames, may allow modern systems to reconstruct the hidden text. Evidence of this includes the developers already mentioned, who report that certain models have successfully decoded messages after more detailed instructions or frame-by-frame analysis. This situation underscores the ongoing struggle between human ingenuity and technological progress.

Expanding the context: the history of visual information hiding

The history of hiding information goes back to ancient times. Long before computers and artificial intelligence, people used a variety of methods to transmit secret messages. One of the best-known examples is steganography — the art of hiding information in such a way that the very fact of its existence remains unnoticed. This could involve writing with invisible ink, concealing notes inside objects, or even using special encryption methods that made text unreadable without a key.

In the age of digital technology, steganography has taken on new forms. Information can be hidden in image metadata, audio files, or even visual elements, as in the case of Ghost Font. An important difference between Ghost Font and classic steganographic methods is that it does not hide information as such; instead, it makes it “invisible” to certain recognition methods. The text exists, it is visualized, but its perception depends on the recognition capabilities of both the human brain and computer systems.

The market and the future: AI and humans in dialogue

The development of Ghost Font points to a growing disparity in perception between humans and artificial intelligence. While computer vision becomes increasingly sophisticated at recognizing static images, the human brain remains unmatched in interpreting dynamic scenes and patterns. This difference could have significant implications for a wide range of fields, from security and monitoring to art and communication.

For example, in cybersecurity, where AI is used to detect threats, Ghost Font could become a tool for hiding malicious code or attackers’ communications. Conversely, developers could use it to create systems that allow only certain types of messages through, filtering out automated AI requests.

In art, Ghost Font opens up new horizons for interactive installations and digital art, where viewers can engage with the work by uncovering hidden layers of meaning. This gives artworks additional depth and intrigue, inviting a more active form of immersion.

However, the development of such technologies also highlights the need for the continual improvement of artificial intelligence algorithms. AI developers are working on models that better understand context, motion, and dynamics, trying to bring their capabilities closer to those of humans. This suggests that the future will likely be an era of gradual convergence between human and artificial intelligence, where both learn from one another. Ghost Font is just one step along this path, illustrating how complex and multifaceted the process of perception and information processing really is.