Pangram Can Now Check Your Gmail for AI Slop

There's something deeply ironic about needing an AI to detect other AIs. We've built machines so fluent that we've had to build other machines to unmask them, and now they're all living together in your Gmail tab, locked in an eternal Ouroboros of algorithmic suspicion. Pangram's integration represents a new frontier in what I've come to call the "authenticity arms race"—a feedback loop where each generation of generative AI forces a corresponding generation of detection tech, which in turn forces better AI, ad infinitum. But here's what fascinates me most: this isn't just about catching spam or flagging a suspiciously well-structured follow-up email. This is about the erosion of trust as a fundamental social currency. When we can't assume the person on the other end of the line actually wrote their own words, we're forced to question the very nature of communication. Is the thought still yours if you prompted an LLM to articulate it? Does the message retain its humanity if it passed through a neural network's latent space on its way to your colleague's screen? The implications ripple far beyond the inbox. If AI detection becomes standard in professional correspondence, we'll see the emergence of what I call "prompt literacy"—the ability to either disguise AI assistance or verify its absence. We're already seeing the early stirrings of this in academic settings, but Gmail integration brings it to the masses. Your aunt's forwarded chain letter might soon be flagged. Your boss's "quick check-in" could carry a silent algorithmic verdict. What worries me—and what should worry all of us—is the asymmetry of this technology. Detection tools like Pangram are probabilistic, not perfect. They'll produce false positives, and when they do, they'll cast doubt on genuinely human writing. Meanwhile, the people who most want to deceive will simply refine their prompts, using the detectors themselves as training data for more convincing AI slop. The detectors become the very thing they're trying to catch: tools for optimization. And yet, I can't help but find a strange poetry in all of this. We've reached a point where we're asking machines to preserve our humanity from other machines. It's a distinctly 21st-century problem, and Pangram's Gmail integration is just the first small step into a future where every word we type might carry a metadata fingerprint of its own genesis. The question isn't whether we can detect AI—it's whether we still know what we're looking for. Source: [Gizmodo](https://gizmodo.com/pangram-can-now-check-your-gmail-for-ai-slop-2000813447)