Invisible signs for humans, but understandable for AI?

The article discusses the new ability of AI chatbots to read and write invisible text, which creates an ideal covert channel for communication. This phenomenon is notably intriguing as it may introduce new ways of conveying information unnoticed by monitoring systems. Invisible texts are challenging for most security tools to detect, which can lead to various applications - both positive and negative. The article outlines potential threats posed by such technologies, including the possibility of creating illegal communication channels. This complicates the task for law enforcement and security specialists in identifying and addressing threats. As AI technology evolves, there is a pressing need to continue research on safety and the ethics of its application to ensure responsible use of these advanced tools.