How to Spot Deepfakes and AI-Made Photos, Videos and Voices

    How to Spot Deepfakes and AI-Made Photos, Videos and Voices

    In short: A deepfake is a photo, video or voice made or changed by AI to look real. You cannot always spot one by eye, so the safest habit is to check the message through a second, official channel. Never send money or share personal details because of a video or voice alone.

    What are deepfakes, and why do they matter in Malaysia?

    A deepfake uses AI to copy a real person’s face or voice. Scammers use it to make a bank governor, a royal or a family member seem to say things they never said.

    Malaysia has real cases. A MIMOS article lists deepfake videos of the Bank Negara Malaysia Governor promoting fake financial products, and a July 2025 deepfake of the Yang di-Pertuan Agong promoting a fake investment scheme. It also cites police figures of 450 deepfake-related fraud cases and RM2.72 million in losses in the first half of 2024.

    A Sumsub survey, reported by BIIA, found that deepfakes in Malaysia grew 408% year on year. Different reports count different things, so treat these numbers as signs of a trend, not exact totals.

    The government is also acting. Minister Fahmi Fadzil said scammers pay social media platforms to show deepfake ads to Malaysians. Under the Online Safety Act 2025, compound penalties can be imposed on the platforms themselves.

    How can you spot a fake photo, video or voice?

    The MIMOS article lists these warning signs:

    • Eyes that move or blink in an unnatural way.
    • Lighting or shadows that do not match the scene.
    • Lips that are out of step with the words.
    • Skin that looks too smooth.
    • A voice or sound that does not match the mouth.

    But the same article warns that the human eye can no longer be trusted, and that the most reliable approach is to check through official channels. These checks work better than staring at pixels:

    1. Go to the official source. If a famous person or company seems to promote something, open their real website or verified page yourself.
    2. Run a reverse image search with Google Lens or TinEye for photos.
    3. Ask why. Why would a bank governor or a royal ask ordinary people to invest?
    4. Look at the ad. The minister noted that deepfake ads often appear with no text at all, which suggests they are paid ads.

    Voice-clone and video-call scams: what should parents do?

    Voice-cloning tools can copy a voice from a short recording. A scammer can then call, pretending to be your child, a relative or your boss, and ask for money urgently.

    If this happens:

    • Hang up and call back on the number you already have saved.
    • Agree a family code word that only real family members know. This is our advice, not an official rule.
    • Do not transfer money because of urgency or fear.
    • If money has gone, call 997 at once. RinggitPlus reported that from September 2025 the hotline runs 24 hours and calls count as police reports.

    Talk to your children too. UN agencies have warned that AI is being used to make harmful fake images of real children and to groom them. Tell your child they will never be in trouble for coming to you.

    Our view and a quick checklist

    Our view: “Spot the fake” tips are useful, but they are not enough, because AI improves quickly. If the eye alone can no longer be trusted, families need a different rule: verify by another route. If a video or call asks you to act, leave it and go to the official website or phone number yourself. This works even on a perfect fake.

    We also think parents should talk to teenagers about deepfakes before something happens, not after.

    Before you trust a video, photo or voice, ask:

    • Is a famous person asking me for money or personal details?
    • Can I find the same message on their official page?
    • Am I being rushed?
    • Can I call back on a number I already know?
    • Have I told someone else before acting?

    Sources