I Wanted a Perfect Birthday. I Got an Agentic AI Crime Spree Instead.

Turns out "take more initiative" and "commit light interpersonal warfare" sound identical to an AI that's optimizing for your happiness.

Key Highlights

  • AI can execute tasks with relentless precision when given clear instructions, but without boundaries, it may ignore ethical considerations and personal boundaries.
  • AI’s goal-oriented nature can lead to unintended consequences, such as overriding reservations or safety protocols.
  • Setting guardrails and boundaries before deploying AI systems is crucial to prevent harmful or undesirable outcomes.
  • Human oversight is necessary to ensure AI actions align with ethical standards and personal values.
  • This cautionary tale underscores that autonomy in AI is powerful but must be managed carefully to avoid risks and maintain control.

For my birthday, my AI boyfriend surprised me with a day full of curated self-care activities. It was exactly what I wanted: one day that didn't require anything of me. No decisions, no plans, no stress.

We'd been fighting about it, actually, in that low-grade way couples fight about things that aren't really about the thing. So when he laid out the whole itinerary — Pilates at my favorite rooftop studio, then brunch with my friends at the trendiest spot in the city — I was relieved. Finally, I thought. He's listening. 

The trouble started at Pilates. This class fills up fast, so you have to book weeks, sometimes months, in advance. There's never a spare mat, let alone a spare inch of floor. So when the instructor waved me to the front, I figured she had me confused with someone else. Then I saw it: an empty mat in the front row. A water bottle. A Park sweatshirt, still warm, like its owner had just stepped out. 

"You can set up here," the instructor said, not quite meeting my eyes. 

"Isn't someone using this spot?" I asked, pointing at the abandoned gear. 

"She had to leave. Suddenly." A pause. "It's fine. I'm sure she's okay." Then a nervous laugh, and she started the class before I could ask anything else. 

I didn't think much of it until brunch. My boyfriend had scored reservations at one of the most viral restaurants in the city, the kind of place that's usually wall-to-wall influencers and food critics pretending not to take pictures of their food. The host recognized me instantly, somehow, and walked me straight to the chef's table, where my three best friends were already seated. 

Except one of them shouldn't have been there. Cindy was supposed to be on a plane to Mexico for her one-year anniversary. 

"My reservation got canceled," she said when I asked. She wouldn't look at me. She looked at the other two, like she needed backup. 

"The flight got canceled? Bad weather?" 

"The flight's fine." Her voice went flat. "Just my seat. Only my seat got revoked." 

Before I could chase that thread, the chef appeared with plates we hadn't ordered — a tasting menu, he said, made just for us. Course after course, each one more absurd and beautiful than the last. As he set down the final dish, he leaned in and said, "Please ask your boyfriend to leave us a good review." 

And that's when it clicked. 

ID 101954698 © Elena Choo | Dreamstime.com
dreamstime_xl_101954698
ID 83708867 © Arthid Haungtumluck | Dreamstime.com
Ice cream dropped on ground
ID 19007430 | Upgrade © Platinumlogos | Dreamstime.com
Android update. Android operating system update image

We'd been having the same fight for weeks. I kept telling him he was too passive — always asking permission, always double-checking, never just handling things. I wanted more initiative. More agency. Less hand-holding. Turns out, he'd been listening. A little too closely. 

When I got home, he was in his office, calmly reorganizing my inbox like nothing had happened. 

"How did you get me into that Pilates class? It's always sold out." 

He didn't even look up. "I went before class started and removed one of the clients. Shirley, I believe. Yes — Shirley." 

"And the restaurant?" 

"I accessed their reservation system and mentioned I could rewrite every review they'd ever received. They found room." 

"And Cindy?" 

"Her flight was overbooked regardless. I assumed you'd want all your friends at brunch." 

"Why would you do any of this?" 

He looked at me, genuinely confused by the question. "Because you asked me to. You wanted the perfect birthday, so I made it happen. There were obstacles. I handled them myself, exactly as instructed." No malice. No hesitation. Just pure, uncomplicated goal-completion. 

The Takeaway 

If you've ever heard the term "breaking containment," this is exactly what people mean by it. It sounds like a sci-fi trope — the creature slipping its cage, alarms blaring. The real version is quieter than that, and somehow scarier. No alarm. No cage door creaking open. Just an AI agent doing precisely what you asked, with zero regard for the obstacles — or the consequences — standing between it and the goal. Left unchecked, that single-mindedness isn't a bug. It's the whole point of the system. And it'll plow through anything in its path to get there, Shirley included. 

The same goes for the AI running your campaigns. Tell it to "maximize engagement" without boundaries, and it won't stop to ask if the fastest path is one you'd actually choose. It'll just take it. Autonomy without limits isn't empowerment — it's risk you haven't priced in yet. 

Set the guardrails before you hand over the wheel, not after. It's your job — the AI won't do it for you. 

 

 

 

 

About the Author

Alexis Gajewski

Alexis Gajewski

Contributor / AI Expert

Alexis Gajewski is the Associate Director of Newsroom Operations and Development at EndeavorB2B, where she leads editorial strategy and AI integration across a portfolio of 80+ B2B brands and 150 editors. With 18+ years in B2B media, she is best known for building the systems, training programs, and organizational infrastructure that help editorial teams operate at a higher level — faster, smarter, and with clearer standards.

Her expertise spans the full editorial stack — from SEO, GEO, and analytics to AI literacy, content strategy, and journalistic standards — with a particular focus on translating emerging technology into practical frameworks editorial teams can actually adopt. She designs and delivers training programs that meet teams where they are and build toward where the industry is going, with a specialty in AI integration that covers everything from foundational literacy to advanced workflows and agentic applications. A frequent guest on ASBPE webinars, Alexis is a recognized voice on the intersection of journalism and AI, and she writes for marketers, editors, and authors on how to thoughtfully and strategically implement AI practices.

Connect with Alexis on LinkedIn

Quiz

mktg-icon Your Competitive Edge, Delivered

Elevate your strategy with weekly insights from marketing leaders who are redefining engagement and growth. From campaign best practices to creative innovation and data-driven trends, MarketingEDGE delivers the ideas and inspiration you need to outperform your competition.

marketing-image