AI can now clone voices and create deepfake videos in as little as 30 seconds, making it easier for scammers to impersonate family, friends, or even officials. A short audio or video clip is all they need to fake a convincing call, message, or FaceTime. (1) But there’s a simple way to fight back, use a personal memory question.
Scammers use AI to mimic the voice or appearance of someone you trust. You might get a call or even a video FaceTime from a loved one saying they’re in trouble and need money immediately. Because their voice sounds real, you react fast, before realizing it’s a scam.
Why Safe Words Aren’t Enough – Use Personal Memories Instead
Safe words are often recommended as a defence against AI scams, but in reality, there are some challenges to this strategy:
1. Under Pressure, People Forget Safe Words
In high-stress situations, like receiving a call from a supposed loved one in danger, people may forget to ask for the safe word or doubt their memory. Panic overrides logic.
2. Safe Words Can Be Compromised
Once a safe word is shared, it can be leaked, whether through accidental disclosure, hacking, or social engineering. If a scammer gains access, the safe word is no longer a safeguard.
3. AI Can Learn Safe Words
If your safe word appears in text messages, emails, or conversations captured in a data breach, AI tools can extract and use it against you. Scammers are getting smarter about finding personal information online.
Instead of relying on a predetermined safe word, ask the caller about something only the real person would remember such as:
- “What did we eat at Grandma’s house last week?”
- “What nickname did I call you when we were kids?”
- “What was the last thing you texted me about?”
This approach works better because it leverages personal memories, which are far more secure than traditional safe words. Unlike a predetermined word or phrase that could potentially be discovered in a data breach, a unique personal memory is not something a scammer can easily guess or uncover through stolen information. Additionally, artificial intelligence, even with advanced voice-cloning capabilities, struggles with real-time recall and the nuances of spontaneous, unscripted conversations.
AI-generated voices can mimic speech patterns, but they cannot convincingly discuss personal experiences in a way that feels natural and accurate. Furthermore, this method is more reliable because personal memories are deeply ingrained in our minds, making them much harder to forget compared to a safe word, which requires conscious effort to recall and use in the right moment. By relying on authentic recollections instead of a static password-like phrase, this strategy enhances security while remaining practical and easy to implement in real-life situations.
In today’s onlife world where AI can clone voices and create deepfake videos with alarming speed and accuracy, it’s crucial to adopt smarter defences against these AI scams. Relying on safe words alone is no longer sufficient, as they can be compromised through data breaches, social engineering, or simple human error. Instead, using a personal memory question provides a far more secure and practical way to verify a caller’s identity.
Unlike safe words, personal memories cannot be easily guessed, extracted by AI, or forgotten under pressure. They tap into unique, shared experiences that only the real person would know, making it nearly impossible for scammers to fake a convincing response.
As AI-driven scams become more sophisticated, we must evolve our security measures accordingly. By incorporating personal memory questions into our verification process, we can better outsmart fraudsters and protect ourselves and our loved ones from deception.
Stay vigilant, question unexpected calls, and always verify before acting, because in today’s onlife world, trust should be earned, not assumed.
Digital Food For Thought
The White Hatter
Facts Not Fear, Facts Not Emotions, Enlighten Not Frighten, Know Tech Not No Tech
Reference:
1/ https://thewhitehatter.ca/the-new-use-of-artificial-intelligence-ai-to-commit-crime/
