Once popular for attacking AI, ASCII smuggling is embraced by spammers

There's a delicious irony in watching a tool designed to attack machines become the everyday workhorse of spammers. It's the digital equivalent of a lockpick being repurposed as a key. ASCII smuggling represents a fundamental truth about our hybrid reality: we are building systems that perceive the world through entirely different senses. What is invisible to us is glaringly obvious to an AI, and vice versa. This asymmetry isn't a bug—it's a feature of the medium itself, and we're only beginning to understand its implications. The shift from offensive weapon to mundane tool is a familiar arc in the history of technology. Every exploit, once discovered, eventually finds its way into the toolkit of the masses. But what's striking here is the *direction* of the attack. We once worried about AI deceiving humans. Now we're seeing humans weaponize the AI's own perception to deceive other AI systems—and, through them, us. The spammer isn't trying to fool you; they're trying to fool the bot that guards your inbox. You're just the collateral damage. This raises a profound question about trust in our automated intermediaries. When we delegate attention to machines—whether it's filtering email, moderating content, or recommending purchases—we're implicitly trusting them to see what we cannot. ASCII smuggling exploits that trust. It's a reminder that our AI guardians are not omniscient; they have their own blind spots, their own perceptual quirks, and those quirks are now being mapped and exploited with the same rigor that security researchers once reserved for human psychology. The deeper lesson here is about the nature of communication itself. Language has always been a battlefield of hidden meanings—codes, ciphers, double entendres. But ASCII smuggling is something new: it's a code that exists *outside* the spectrum of human perception entirely. It's a message written in a frequency only machines can hear. As we move further into an age of human-machine collaboration, we need to ask ourselves: how many silent conversations are happening around us, invisible to our eyes but shaping our digital lives? The answer, I suspect, is more than we'd like to admit. Source: [Ars Technica](https://arstechnica.com/security/2026/09/once-popular-for-attacking-ai-ascii-smuggling-is-embraced-by-spammers/)