Kids IT Lab No.15 | Security

Cybercrime in the AI Era

AI can now generate natural-sounding text, voices that mimic real people, and realistic images and video. That is a huge benefit for creativity — but it also makes it far easier to run scams and impersonation attacks. This article explains what has changed, and how to defend yourself by verifying through a second channel rather than trying to spot a fake.

What Has AI Changed?

In the past, you could often spot a scam email because the wording sounded unnatural or the images were low quality. Generative AI has changed that. Natural-sounding messages in any language, realistic images, and voice cloning that mimics a specific person are now accessible tools. On top of that, deepfake technology can splice a celebrity's or family member's face into someone else's video. The era of "I can trust my eyes and ears" is ending.

AI-Powered Crime Tactics

How AI Upgraded Crime: Before vs. Now "Spotting bad grammar" no longer works as a detection method Tactic Before (~2022) AI era (now) Phishing email Fake site redirect Awkward wording "Dear valued customer..." Perfect language + personalized ChatGPT-generated with your name Voice impersonation Phone call as a family member Different voice gave it away Basic grandparent scam level Perfect clone from 3 seconds of audio Tools like ElevenLabs available to anyone Video / face spoofing Video calls, social media Sloppy image edits Required specialist skills Real-time deepfake video One SNS photo → animated face Spear phishing Target-specific attacks Weeks of manual research Handwritten custom messages AI auto-scrapes your SNS profile Personalized messages in minutes Malware creation Attack software Required advanced coding Pattern-based detection worked AI generates unique code every time Pattern detection less effective ▶ "Spotting by appearance" is no longer reliable. "Verify by channel" and "confirm with a passphrase" are the new baseline. Agree on a passphrase with family and friends in advance — it exposes voice and face spoofing instantly.
Figure 1: How AI has upgraded crime. Quality of fakes has jumped — "looks real" is no longer enough to trust.

How Deepfake Scams Work

New Defense Basics Against AI Crime: Channel and Passphrase Verification Voice and video can no longer confirm identity. Agree on a method with family and friends in advance. Verify by Channel ▶ Confirm via a different channel than the one used to contact you DM → call; phone call → meet in person ▶ Call back using a number you already know Use your own contacts — not a number given by the caller ▶ Anyone rushing you is a red flag "Send it right now" is a classic crime signal ▶ Put unknown contact on hold No matter how urgent it sounds, take 5 minutes A real person can always be reached another way Confirm with a Passphrase ▶ Agree on one phrase with your family Your pet's name, a shared travel memory, etc. ▶ Use it on a call to verify identity Ask "What's the passphrase?" — if they know it, you're good ▶ Do not hint at it on social media If you post your pet's name or birthday, pick something else ▶ Works with friends too Instantly tells you if a DM is from the real person or a hijacked account Something AI can never fake
Figure 2: Two new defense basics against AI crime. Agreeing on a family passphrase in advance is the most powerful countermeasure.

People who post lots of video and audio content on social media are at greater risk of becoming impersonation targets. Criminals combine photos, short audio clips, school names, club activities, and relationships to build convincing messages. That is why checking whether a voice or face matches is no longer enough — you must also evaluate the channel a message came through and whether the content feels off.

Situations Where Teens Are Most at Risk

The scenarios most relevant to teens are: social media DMs, in-game chats, resale app messages, video comment sections, and fake giveaway campaigns. The hook is always urgency, fear, or the promise of gain: "Grab this limited item," "Confirm your identity by logging in here," or "Contact me if you want this video taken down." As AI makes the wording more natural, it becomes harder to notice the scam at first glance.

A DM using a friend's name does not mean the message is from your friend. Their account may have been hijacked, or it could be a fake account with a nearly identical name. If someone asks for money, a password, a verification code, your address, or photos of yourself, do not reply right away — verify through another channel first.

How to Protect Yourself from AI Attacks

Spotting AI-generated fakes by appearance or voice alone is increasingly unreliable. The new baseline is: verify using channel and passphrase. Agree on one passphrase with your family (to use when verifying identity over a phone call). If you receive a message from an unusual account, confirm with the person through a separate channel. If money comes up, pause — re-confirm in person or via another method before doing anything. These habits make most impersonation scams much harder to pull off.

Common Pitfalls

Misconceptions About AI Crime
  • Thinking "I'm not a celebrity so nobody will use my face." If your face is on social media, you are already a potential target.
  • Believing "video call = verified identity." Real-time deepfake technology is already in practical use.
  • Overconfidence that "you can spot AI by skin texture." The technology advances on a scale of months, not years.

How Will This Help You in the Future?

Fighting AI-powered crime also requires AI-powered defense — and engineers who build it. Deepfake detection, spam filtering, anomalous login detection, and identity verification systems are all active research and product areas. Teens who develop exposure to both AI and security have a clear path toward contributing to this field. AI crime is also a major topic in digital literacy education.

Take Action Today

Start with 3 steps
  1. Agree on a passphrase with your family and make a plan to use it when someone calls about an urgent money situation.
  2. Review how much video and audio you post publicly on social media, and reduce unnecessary material.
  3. If a suspicious video or message arrives, verify the person through a second channel (in person or by phone) before responding.

Summary

AI has made it hard to detect impersonation by appearance or voice alone. The new baseline is verifying through a separate channel and confirming with a passphrase. Limit the raw material you put online, and agree on a family passphrase ahead of time. Defending against AI crime is itself an exciting and growing career path.

Check Fraud in the AI era grows as?

Related articles

PreviousOnline Crime Methods to Watch Out For NextStaying Safe on School PCs RelatedWhat Is ChatGPT? A Beginner's Guide for Teens RelatedHow to Use AI Chat Safely
← Back to Security