Information Security Awareness

Jump to a Security Awareness page:

Don’t get caught in the deep end!

Watch out for deepfakes!

What is a deepfake?

A deepfake is a fake audio file or recording, image or video generated by machine learning (aka deep learning).

It is a type of artificial intelligence (AI) designed to replace a person’s likeness in a convincing way.

Can deepfakes be used for legitimate purposes?

There are legitimate use cases of deepfakes, which include, but are not limited to:

• Call forwarding and receptionist services,
• Customer support,
• Help people who have lost their voice speak again through an avatar.
• Provide anonymity for people providing testimony in public trials, or for journalists’ sources.
• Recreating a crime scene to help with an investigation.
• Creating an image in your likeness to use to model clothing on fashion sites to see how they would fit and look, and if you’re interested in purchasing them, and
• Speed up video game creation by creating environments, sounds, and other components.

What are examples of malicious uses of deepfakes?

Cybercriminals can create deepfake images and videos for various purposes, including, but not limited to:

• Creating false narratives about public figures and individuals.
• Impersonating public figures so they appear to encourage people to get involved in a scheme that results in the cybercriminals conducting fraud.
• Attempting reputational damage of the person who is being deepfaked.

How does someone spot a deepfake?

• In audio:
  • Listen for unnatural pauses, and robotic-sounding speech.
• In images:
  • Unnatural skin tone colouring.
  • Unrealistic-looking hair or teeth.
• In videos:
  • Lack of the subject blinking. Deepfakes don’t blink.
  • No side profiles.
  • Voice intonation and cadence that sound different from how people normally talk.
  • Poor-quality images and video.
  • Video speed is faster than human norms.
  • Difference in colours and lighting.

Take the Deepfakes Quiz!