How to protect yourself from AI-powered attacks

Scams have always existed. In the past, criminals would knock on doors posing as bank employees. Today, they use artificial intelligence to send audio messages with the voice of a family member asking for money. Technology has amplified fraud, making it more realistic, but those who understand the strategies used by scammers have a better chance of avoiding them.

Leonardo Oda, a marketing and technology specialist, warns that AI has changed the game. "Criminals exploit fear and urgency to get victims to act without thinking. Therefore, the best way to protect yourself is to recognise the signs and adopt extra checks before making any decisions," he says.

Audio and Video Scams: Be wary of your own perception.

AI voice cloning has reached a disturbing level. Criminals now need only a few seconds of audio to precisely replicate someone's voice. This means a simple WhatsApp message or social media video could be enough to create convincing fake audio recordings that deceive even close family members.

To protect yourself, Oda recommends confirming the identity of the person before taking any action. If you receive a money request by audio, call the contact back and verify the information. If the person doesn't answer or responds with vague phrases, like "I can't talk now," increase your suspicion. Another effective measure is to agree on a secret keyword with family members to be used in emergencies.

It's also important to pay attention to details that might expose fraud. Unusual pauses, a slightly distorted tone of voice, or a mechanical speech rhythm are signs of AI manipulation.

In video cases, deepfake technology makes it even harder to distinguish what is real from what has been manipulated. Misaligned lip movements with the audio, artificial facial expressions, irregular blinking, and distortions in the background of the image can indicate a forgery. Furthermore, the voice may exhibit unusual intonation or robotic pauses.

The best defence is to be suspicious, even if it seems genuine. Fraudsters are using AI to manipulate emotions and create a sense of urgency. If something seems odd, check it out before acting," Oda stresses.

Banks and businesses don't ask for data via text message.

"Messages supposedly sent by banks or companies, warning of urgent account blocks, are another common tactic," warns Oda. Using AI, fraudsters can create convincing emails and text messages, mimicking official communications to trick victims into clicking on fraudulent links.

To avoid this type of scam, the advice is never to click on links received via email, SMS, or WhatsApp. Ideally, check the bank's official website directly or contact the telephone number listed on your card. It's also important never to share passwords or verification codes, and to be wary of alarming messages, as criminals exploit fear to get victims to make hasty decisions.

False promotions and non-existent prize draws

Congratulations! You've won a brand new car, just pay a fee to release it." Who hasn't received a message like that? With AI, these scams have become even more sophisticated. Criminals are creating manipulated videos and audio recordings, using the faces and voices of celebrities to promote fake prizes, making the scam even more believable.

"The first step to protecting yourself is remembering that if you didn't sign up for a promotion, there's no reason to win anything," says Oda. Furthermore, overly enticing offers should be viewed with suspicion. Before believing the promise, check the company's official website to verify if the promotion actually exists. And under no circumstances should you provide bank details to release a prize.

Information and caution are the best defence.

Fraudsters exploit misinformation and urgency to deceive victims. Leonardo Oda recommends sharing these guidelines with family members, especially the elderly and teenagers, who are often frequent targets.

The best way to avoid a scam is to slow down. If something seems odd, stop, think, and confirm before making any decisions," concludes the expert.