Deepfake scams are getting more convincing, but some old-school tricks still do the job. Companies are using low-tech moves like asking callers to draw a smiley face or hold it up to the camera. Other tactics include prompting the caller to move the webcam around or asking questions only a real colleague would know. If suspicions rise, hanging up and calling back on a trusted number is a quick way to verify authenticity. These simple methods are surprisingly effective right now and can catch scammers off guard.

Why Human Checks Still Matter in Detecting Fakes

Many security leaders say that blending basic human challenges with policies works best. It’s not just about relying on detection software, which can be fooled. Instead, combining simple social engineering tactics with layered checks makes scams much harder. For example, asking someone to do a quick doodle or move the camera can reveal if they’re real or fake. This approach recognizes that social engineering—exploiting human behavior—is still the biggest threat, even with advanced AI tools in play.

Deepfake fraud losses topped $200 million in just the first quarter of 2025. That’s why traditional firms are testing procedures like callback protocols and passphrases. These methods act as cross-checks, making it harder for scammers to succeed. Experts recommend using multiple layers of verification, including checking the origin of images or videos. It’s not about one perfect detector but about creating a workflow that combines different safeguards to spot fakes more reliably.

Provenance and the Role of Technology in Verifying Media

This week, Google announced that its Pixel 10 phone and Photos app will include C2PA Content Credentials. These digital “nutrition labels” carry cryptographic info about how images and videos were made. It’s a way to prove the origin, not just detect fakes. When platforms show provenance info clearly, it helps users and organizations trust what they see. If everyone adopts these standards, it will become normal to verify media instead of just relying on detection tools alone.

Law enforcement is also improving at recovering money from scams. Earlier this year, Italian police froze nearly €1 million from an AI voice scam that impersonated a government minister. It wasn’t perfect, but it was a step forward. Meanwhile, simple physical checks can block scammers in real time. For instance, a finance manager once stopped a scammer by asking the fake CFO to angle the webcam toward a whiteboard. The lag and awkward silence gave away the scam, showing that basic human checks still work. Techniques like changing lighting or holding up a newspaper can disrupt pre-recorded videos and reveal fakes.

These tactics are supported by policies and standards. Platforms need to display provenance info and organizations should integrate checks into their workflows. YouTube’s move to label unaltered clips with Content Credentials hints at a future where verification is built into media sharing. Detection tools are not going away but are moving to the background. Instead, front-line defenses are about verifying identity and authenticity before reacting.

For small teams worried about overkill, simple steps are enough. Pick two easy-to-learn methods, like using weekly-changing passphrases or calling back on a trusted number for money requests. Tape reminders nearby to reinforce these habits. It’s not glamorous, but it’s much better than losing millions to a fake Zoom call. The key is encouraging people to slow down and double-check—this “pause” can stop many scams in their tracks.

In the end, asking for a smiley face or performing a quick verification isn’t a joke—it’s a smart pattern interrupt. When combined with provenance checks and a culture that emphasizes caution over haste, these tactics give organizations a real edge against increasingly convincing AI fakes. It’s about mixing old-school human tricks with new technology to create a balanced defense that’s harder for scammers to crack.

Inspired by

• https://ai2people.com/draw-me-a-smiley-face-old-school-tests-are-tripping-up-new-school-deepfakes/