AI “epic fails” aren’t just funny headlines; they’re patterns you can design against. In this episode of Pop Goes the Stack, #F5's Lori MacVittie, Joel Moses, and Buu Lam walk through why so many AI-powered features keep going off the rails, from chatbots inventing policies to agents deleting real infrastructure, and what those failures teach us about building safer systems. Joel frames most incidents in two buckets. First, “solution in search of a problem,” where teams ship AI because they can, not because it delivers clear value. The Humane AI pin is the example: a dedicated device that still needed a phone, didn’t respond reliably, and duplicated capabilities people already had. Second, treating a statistical prediction engine like an authority. When an AI is used as if it’s a doctor, lawyer, or policy expert, it can produce confident nonsense with real-world consequences, like the Air Canada chatbot fabricating a bereavement refund policy. Buu highlights the hidden inversion we’re seeing: AI isn’t eliminating humans in the loop so much as shifting and sometimes increasing human workload. Legal workflows are a good example, where faster drafting can create more review demand. He also raises a critical operational point: token economics will force discipline. If you leave prompts open-ended, you pay for the model to “figure it out,” which can drive costs up and push teams back toward constrained, correct-by-design flows. The practical enterprise takeaway is permissions and agency. An agent “doing the thing” is still doing it as you. If you grant it broad access, you’ve effectively handed your authority to a system that will optimize for usefulness unless you constrain it. Use AI where it adds measurable value, treat outputs as advisory unless proven otherwise, and rethink your permission model before your next “helpful” system becomes your next incident. Chapters: 00:00 Welcome to Pop Goes the Stack 01:05 AI epic fails: When “innovation” goes off the rails 02:00 Fail category #1: Solutions looking for problems 02:41 Fail category #2: Treating statistical prediction engine like an authority 03:17 Humane pin fail: Needs a phone, adds surveillance, no value 05:01 McDonald’s milkshake AI reverse-engineering (and shutdown) 06:25 Failing while “working”: Misaligned value and incentives 06:54 Utility vs correctness: The core mismatch in AI systems 07:44 Legal AI irony: More AI → more lawyers (human in the loop persists) 08:59 “AI runs DNS?” and why hallucinated authority is dangerous 10:21 Will 2026 be the year of what database did AI delete today? 11:03 Exam proctoring fail: “Detecting deviation,” not cheating 11:45 Humans are edge cases: Bias becomes harassment at scale 13:51 Air Canada: How much is train AI vs train the people? 16:08 Token economics: Prescriptive scenarios, sessions, and usage-based subscriptions 17:57 Key takeaways: Selective AI-enablement + value-first + treat agent actions as "you" Want to dive into other AI fails, read the article: https://go.f5.net/1b1sgzyz Learn how you can stay ahead of the curve and keep your stack whole with additional insights on app security, multicloud, AI, and emerging tech: https://go.f5.net/kw9ci6o1 More about F5: https://go.f5.net/u4as6t45 Read our blog: https://go.f5.net/zgvqdghf Follow us on LinkedIn: https://go.f5.net/hfh46t03

 46      4