AI chatbots can read and write invisible text, creating an ideal covert channel

https://arstechnica.com/security/2024/10/ai-chatbots-can-read-and-write-invisible-text-creating-an-ideal-covert-channel/

It’s like one of those videos of early computer programmers waiting with their punchcards. Nothing is agreed until he’s run it past his AI. You’d think you get to be a senior because you have judgement, you’re capable of making a decision and standing by it. You’re decisive… not *it* is decisive. So he demands our ideas are “fit for AI”. No sideways thinking, no weirdness, “inside the box”, safe in: safe out. Hybrid efficiency. Augmented intelligence. So we’ve made our best thoughts invisible. Like kids passing lemon juice written notes in class, we sneak weird prompts in and wait.