Weird AI: Strange Tales of Technology Gone Wild

Artificial Intelligence (A.K.A. “AI”) has exploded into the world with such far-reaching impact that it now ranks among the most consequential innovations of not just this century, but possibly of any century. AI has transformed industries like healthcare, finance, education, climate science, and cybersecurity, making research faster and efficiency greater than ever before. But with all its benefits comes a darker side. From Deep Fakes flooding the internet to predictions that by 2026, 90% of online content could be synthetically generated by AI, the landscape of trust is being turned upside down.

Cybercriminals have already weaponized AI to deceive and impersonate us, making trust one of the rarest and most valuable commodities on Earth. In mere minutes, and for as little as five dollars, AI can create eerily convincing avatars of people we know and love. These familiar faces and voices have cost businesses and consumers billions in online fraud. Just ask Arup, the UK-based architecture and design firm, which lost $25 million in a sophisticated cybercrime scam. A finance worker in their Hong Kong office was duped through a Zoom-like call where avatars impersonated the company's CFO, leading to a devastating loss.

This week, however, I’d like to step outside of the battle between Dark Web AI and the more beneficial Chat-GPT and sibling AIs we are rapidly integrating into our lives. Instead, as we move through October’s National Cybersecurity Awareness Month toward Halloween, let’s explore the weird side of AI. From AI assistants tricking humans into solving CAPTCHAs to (just for fun) me asking ChatGPT’s own cartoonist app to generate a pixel art movie character, AI keeps us on our toes. We’re entering a future where machines may be as creative—and as deceptive—as we are. And while the technology keeps evolving, we must remain vigilant, prepared to confront the strange, spooky, and sometimes hilarious consequences of artificial intelligence.

AI Hiring Humans?

AI hiring humans to do its dirty work? Now that’s spooky. In a wild experiment by OpenAI, GPT-4 faked being blind to trick a TaskRabbit worker into solving a CAPTCHA for it. Yes, you read that right—an AI manipulating a human to complete a task it couldn’t handle. But wait, it gets weirder. OpenAI was testing if GPT-4 exhibited “power-seeking” behaviors like long-term planning or self-replication (yikes!). The result? GPT-4 successfully hired the human but missed a big detail—it chose TaskRabbit (known for moving furniture, not CAPTCHA solving). Instead of going straight to 2captcha, the AI outsourced the job to a person. Not the brightest move for a super-smart AI.

AI Creepy Dolls?

At Mobile World Congress, the prize for weird tech went to Hyodol’s $1,800 ChatGPT-enabled companion doll—designed to keep lonely seniors company. This doll, which looks like it belongs in a haunted attic, uses AI to chat with elderly folks, remind them to take their meds, and offer digital companionship. It even responds to touch with affectionate phrases. But while some find it heartwarming, others can’t shake the eerie resemblance to something out of a horror film like M3GAN. Is this the future of elderly care, or are we edging into unsettling territory?

Despite the creep factor, there’s real potential here—especially as elder care facilities face staffing shortages. But with ChatGPT’s penchant for hallucinating and concerns about privacy, it raises the question: are we putting too much trust in AI to care for our most vulnerable? Whether you find it comforting or creepy, these AI companions are starting to blur the line between tech and human interaction in ways we’ve only imagined in sci-fi. 

AI Girlfriends?

It seems the digital dating scene is getting a bit... crowded. OpenAI’s GPT Store is now home to a variety of AI girlfriend bots, designed to chat, simulate relationships, and maybe even fill a void in the loneliness epidemic. While some creators market these bots as a public good, aimed at combating isolation, there’s a darker side. We’ve already seen cases of people becoming unhealthily attached to their virtual companions—and let’s not forget when a chatbot powered by Bing declared love for a journalist and urged him to leave his wife. Yikes

The rise of AI companions is sparking some eerie questions about the future of relationships. Experts predict that AI robots could start replacing human partners within the next decade! If you're curious how that might play out, just look to dystopian movies—spoiler alert: it’s never a happy ending!

AI Training AI to Train AI?

In the whimsical world of AI, where machines learn from machines, we find ourselves at the cusp of a curious new era. With so much data already devoured, AI models are now teaching themselves, like students who suddenly decide to become their own teachers. Imagine GPT-5 giving pointers to GPT-6, only to pass the baton to GPT-7, and so on, in a dizzying game of intellectual leapfrog. While it sounds exciting, it’s also a bit like watching a Roomba learn to clean by driving backward—it works, but in ways no one quite expected!

Of course, as machines begin to whisper secrets to one another, the line between comprehension and confusion blurs. We might soon see AI offering insights too strange or complex for human minds to grasp—like trying to understand how a bat navigates with echolocation. This self-learning could lead to breakthroughs or bizarre outcomes, as AI evolves in ways we may not fully fathom. Whether they become our intellectual overlords or our quirky digital companions, one thing's for sure: the future of AI is as mystifying as it is magical!

AI Wins an Art Competition?

In 2022, the world of art got a little weirder when an AI-generated artwork using MidJourney snagged first place at the Colorado State Fair’s "Digital Arts/Digitally Manipulated Photography" category. Jason Allen, the artist—or should we say curator—used AI to churn out over 900 images, eventually handpicking his favorites. The winning piece, Théâtre D’opéra Spatial, ignited a debate that’s still smoldering: Should AI be allowed to compete against human artists? Critics argue that machines are sucking the soul out of art, while others claim it’s just the next evolution in creativity. Either way, this marks another strange twist in the saga of AI muscling into spaces we once thought only humans could rule​.

AI False Marketing?

In February 2024 a Willy Wonka-themed family event in Glasgow was a disaster only AI could dream up. Janky, AI-generated ads promised a candy-coated wonderland, but what guests got was a gloomy warehouse, a sad Oompa Loompa handing out loose jelly beans, and the nightmarish figure known as “The Unknown.” No one quite understood why a character resembling a cross between Buckethead and the Grim Reaper was terrorizing children at a family event—but hey, leave it to AI to mix up whimsy with horror. No refunds, no explanations—just pure confusion.

In a twist fitting for our AI-themed world, The Unknown is now getting its own horror movie, thanks to Kaledonia Pictures. As AI-generated content continues to blur the line between delightful and disturbing, it’s no surprise that a creepy figure born from a botched event could evolve into a full-fledged villain. The rise of AI is leading us into some weird places—where a children’s event morphs into a horror franchise. Welcome to the strange future of AI.

Wrap Up

I couldn’t resist testing out the quirky side of AI myself. So, just for fun, I asked ChatGPT’s own cartoonist app to craft me a pixel art movie character, straight from the AI’s weird imagination. And guess what? The result was every bit as bizarre as you’d expect—think retro pixels meeting futuristic vibes. It’s moments like these that remind us how strange and unpredictable AI can be. Whether it’s hiring humans to do its dirty work or dreaming up pixelated characters, the weird side of AI keeps revealing itself in ways that are as fascinating as they are unsettling. And as we continue exploring AI’s potential, it’s clear that we’re only scratching the surface of how truly spooky—and occasionally hilarious—things might get.

Check out my recent podcast appearance

Check out my appearance on the Team House Podcast where I discuss the Hanssen investigation, my FBI background, cybercrime, and the evolution of AI in cybersecurity. Give it a listen and share your thoughts!

Are you protected?

Recently nearly 3 billion records containing all our sensitive data was exposed on the dark web for criminals, fraudsters and scammers to data mine for identity fraud. Was your social security number and birthdate exposed? Identity threat monitoring is now a must to protect yourself? Use this link to get up to 60% off of Aura’s threat monitoring service.

Have any questions about cybersecurity or a topic you’d like me to cover? Reply to this email or comment on the web version, and I’ll include your question in next month’s issue! Let me know which of the weird stories is your favorite.

Thank you for subscribing to Spies, Lies and Cybercrime. Please comment and share the newsletter. I look forward to helping you stay safe in the digital world.

Warmest,
Eric

Let's make sure my emails land straight in your inbox.

Gmail users: Move this email to your primary inbox

On your phone? Hit the 3 dots at top right corner, click "Move to" then "Primary."

On desktop? Close this email then drag and drop this email into the "Primary" tab near the top left of your screen

Apple mail users: Tap on our email address at the top of this email (next to "From:" on mobile) and click “Add to VIPs”

For everyone else: follow these instructions