Send a text

Ever wondered what happens when AI tries to scare three sleep-deprived hosts who can’t resist breaking character? We put four machine-written horror stories to the test—reading them aloud, acting them out, and grading them with ruthless honesty. From a backyard camera that says “Hello, homeowner” to a silent second floor that refuses to light up, from a circling shark that turns “calm” into “controlled panic” to a polite phone alert about a gravitational anomaly, the night swings between genuine tension and glorious derailment.

We start with the show’s rotating-judge format and jump straight into the chaos: a security-cam creep that almost works, a haunted-house walk-through that mistakes doors for dread, an ocean encounter where clear staging actually builds momentum, and a cosmic-horror premise with bureaucratic phone notifications that feel eerily plausible. Along the way, we unpack why AI nails setup but fumbles character voice, escalation, and payoff. Horror breathes through choices, stakes, and sensory detail; prompts can sketch the scene, but people have to make it pulse.

There’s plenty of laughter—improv accents, GTA quotes, and a new classic line about “controlled panic”—but we also dig into real craft notes any writer can use. We talk geography in survival horror, how to avoid filler dialogue, when to reveal the monster, and why cosmic dread needs awe plus consequence. The result is part live table read, part writing workshop, and part roast of our robot collaborator.

If you like found footage vibes, urban legends, shark thrillers, or cosmic weirdness, you’ll find a moment that sticks. Hit play, rate the stories with us, and tell us which one you’d rewrite—or which one actually gave you chills. Subscribe, share with a friend who loves horror, and drop your own AI prompts or fixes in a review so we can try them next time.