Skip to Content
View all events

AI Horror Stories

Published by O'Reilly Media, Inc.

Beginner content levelBeginner

When AI takes a terrifying turn

AI is moving fast. So are the ways it can go spectacularly wrong. Vibe-coded apps leaking sensitive data. AI integrations that seemed clever until they weren’t. Or autonomous agents giving you everything you wished for, but that you didn’t realize you were asking for. The scariest AI stories aren’t about machines turning on us. They’re about the very human decisions that set the stage.

Because for every robot with glowing red eyes, there was an engineer who installed those LEDs on purpose.

Join us for an event with O’Reilly authors, practitioners, and industry experts who have seen what happens when AI meets the real world without adequate guardrails, governance, or good judgment. They’ll share their mistakes, the consequences of those mistakes, and what you can actually do to keep your own AI daydreams from becoming nightmares.

This live event is for you because...

  • You’re shipping something with AI assistance and feel uneasy at not being entirely sure what’s actually running in production.
  • You’re on a team that’s integrating AI tooling into real workflows and you want to build the instincts to catch problems before they become incidents.
  • You believe that keeping humans meaningfully in the loop isn’t a limitation of AI—it’s the whole game.

Prerequisites

  • Come with your questions
  • Have a pen and paper handy to capture notes, insights, and inspiration

Recommended follow-up:

Schedule

The time frames are only estimates and may vary according to how the class is progressing.

Alistair Croll – Introduction (5 minutes)

Alistair Croll welcomes you to AI Horror Stories.

When the Attendees Start Moving the Bar – Alistair Croll (25 minutes)

For the last two months, Alistair Croll has been building a conference platform. With a few decades of running conferences, that isn’t surprising; what’s unexpected is that the attendees are agents. And to test the platform, you need AI agent attendees, controlled by AI-driven synthetic humans. Things turn into Inception pretty quickly. Alistair offers examples of the kinds of emergent and sometimes disastrous behavior that happens when you fill a platform with AIs—and ask them to post bios, open booths, propose and rate talks, and socialize.

Publishing My Secrets on GitHub: A Claude Code Story – Dan Shapiro (25 minutes)

What could be more innocuous than filing a bug? Claude Code can do that for you. And while it’s there, it can also publish every secret on your local computer to the open internet. How? It’s easy! Dan Shapiro, pioneer of desktop laser technology, shows you what AI can get up to when you’re not looking.

When AI Makes Progress Look Real – Patrick McFadin (25 minutes)

Have you ever had AI help you move so fast that you started to believe you were unstoppable? The code looked clean, the tests were green, and everything seemed to be falling into place. Patrick McFadin, principal technical strategist at IBM, shares an AI-assisted development story that felt like a breakthrough right up until reality barged in and exposed what was really happening. Using a real project involving Rust and Cassandra SSTables, Patrick shows you how easy it is to mistake motion for progress, how quickly confidence can outrun truth, and why human judgment still matters most when AI seems to be doing everything right. You’ll also hear about the principles Patrick now works by to ensure AI accelerates innovation rather than derailing it.

Alistair Croll – Closing Remarks (5 minutes)

Alistair Croll closes out today’s event.

Your Hosts and Guests

  • Alistair Croll

    Alistair Croll is an entrepreneur, author, and conference organizer. He's written four books on technology and society, including the best-selling Lean Analytics, which has been translated into eight languages. He's the cofounder of web performance startup Coradiant (acquired by BMC), the Year One Labs startup accelerator, and a number of other early-stage companies.

    A prolific speaker, Alistair was a visiting executive at Harvard Business School, where he helped create a course on data science and critical thinking. He's founded and chaired a number of the world's leading technology events including Cloud Connect, Strata, Startupfest, Scaletech, and the FWD50 Digital Government conference. He's currently working on Just Evil Enough, the subversive marketing playbook. Alistair lives in Montreal, Canada, and writes at acroll.substack.com.

  • Dan Shapiro

    Dan Shapiro is an entrepreneur, innovator, and author who’s transforming how people create. He wrote Hot Seat: The Startup CEO Guidebook (O’Reilly), and as CEO and cofounder of Glowforge, he pioneered desktop laser technology. Dan also created the board game Robot Turtles, which teaches programming fundamentals to preschoolers and has been sold everywhere from Target to the Museum of Modern Art. Previously, he was CEO of Google Comparison Inc. (following Google’s acquisition of his startup Sparkbuy) and founder of Photobucket Inc. (formerly Ontela). A prolific innovator with over 100 patents to his name, his work has been featured in the Wall Street Journal, NPR, and on the front page of The New York Times.

  • Patrick McFadin

    Patrick McFadin is a principal technical strategist at IBM, where he focuses on open data architectures for modern applications and AI. A longtime Apache Cassandra committer, ASF member, and data infrastructure strategist, he has spent his career helping engineers build systems that survive real-world scale. Patrick is also coauthor of Managing Cloud Native Data on Kubernetes (O’Reilly) and a frequent speaker on distributed systems, databases, and developer experience.

Skill covered

Artificial Intelligence (AI)