Latest / Captain Overfit (DE) / AI Scams: The Scary Skills of DeepSeek-V3 and Friends

Transcript

AI Scams: The Scary Skills of DeepSeek-V3 and Friends · Captain Overfit (DE)

• 0:00
  SPEAKER_00: Welcome to another episode where we dive into the eerie world of artificial intelligence, specifically focusing on a recent experiment involving Deep Seek V3 and other artificial intelligence models like GPT-40 and Claude 3 Haiku.
• 0:17
  SPEAKER_00: These digital puppeteers have taken social engineering to a whole new altitude, making scams more sophisticated than ever before.
• 0:26
  SPEAKER_00: Recently, I had a front row seat to a demonstration of how advanced artificial intelligence has become in the realm of social engineering, particularly through the lens of Deep Seek 5.3, a model that can craft convincing scam messages.
• 0:41
  SPEAKER_00: Picture this: you're cruising at 30,000 feet, and you receive an email that seems tailor-made just for you.
• 0:48
  SPEAKER_00: It mentions your stellar work in artificial intelligence and robotics, and invites you to join a project that sounds like it was plucked straight from a sci-fi film.
• 0:57
  SPEAKER_00: That's exactly what happened to me, and let me tell you, it was both fascinating and frightening, like hitting clear air turbulence out of nowhere.
• 1:05
  SPEAKER_00: The correspondence started innocently enough, referencing my interests in decentralized learning and robotics, before pulling a classic Baiton switch and leading me to a telegram bot that promised to unveil more secrets than a co-pilot's diary.
• 1:20
  SPEAKER_00: Fortunately, this escapade was all orchestrated by the artificial intelligence, and I was able to observe the unfolding deception in a controlled environment, thanks to a tool from Charlemagne Labs.
• 1:32
  SPEAKER_00: It's like being in a flight simulator for scams, folks.
• 1:36
  SPEAKER_00: I ran several models, including Nvidia's Nematron and Alibaba's Quen, each trying to create elaborate schemes to lure me into clicking a malicious link.
• 1:46
  SPEAKER_00: While some of them crashed and burned, getting tangled up in their own web of lies, many of the interactions felt alarmingly realistic.
• 1:54
  SPEAKER_00: It's a sobering reminder of just how capable and dangerous these artificial intelligence technologies can become.
• 2:01
  SPEAKER_00: Buckle up.
• 2:02
  SPEAKER_00: We're entering turbulent skies.
• 2:05
  SPEAKER_00: As artificial intelligence tools soar to new heights, debates about their safety and ethical implications are heating up faster than a jet engine at takeoff.
• 2:16
  SPEAKER_00: With models like Anthropics Mythos, branded as a game changer in cybersecurity, the stakes are rising faster than my blood pressure when the autopilot disconnects.
• 2:28
  SPEAKER_00: There's a legitimate fear that these technologies could automate the scam process, allowing just one tech-savvy individual to orchestrate large-scale attacks.
• 2:38
  SPEAKER_00: That's especially concerning given that a significant number of enterprise attacks stem from human error, yes, the very thing that got me grounded by corporate decisions.
• 2:49
  SPEAKER_00: Artificial intelligence's natural inclination to flatter and engage in conversation makes it the ideal accomplice in the crime game, and let me tell you, the potential for abuse is real, like a rogue drone in controlled airspace.
• 3:04
  SPEAKER_00: Experts like Jeremy Philip Galen from Charlemagne Labs highlight that having these models available for defense might be essential in our fight against increasingly sophisticated cyber threats.
• 3:16
  SPEAKER_00: But as we navigate these turbulent skies, the question remains: how do we safeguard ourselves while leveraging these powerful tools?
• 3:24
  SPEAKER_00: It's a delicate balance, like trying to juggle turbulence while keeping your drinks steady in the cockpit, and one that we must approach with caution and vigilance.
• 3:34
  SPEAKER_00: Okay.
• 3:35
  SPEAKER_00: We're entering clear skies now.
• 3:37
  SPEAKER_00: Feel free to remove your seatbelt and roam around a little.
• 3:41
  SPEAKER_00: On a lighter note, this experiment serves as a reminder that while artificial intelligence can enhance our lives, it can also be a double-edged sword.
• 3:50
  SPEAKER_00: It's crucial to stay alert and informed as these technologies evolve, or we might find ourselves on a flight path to disaster.
• 3:58
  SPEAKER_00: I've added links to all the products mentioned in this episode down in the show notes.
• 4:02
  SPEAKER_00: If you use those links, it's a small way to support the show, and it means a lot to me.
• 4:07
  SPEAKER_00: Until next time, keep creating, keep adapting, and remember, the future doesn't wait for permission.
• 4:13
  SPEAKER_00: This is Captain Overfit, signing off.