Deepfakes

A deepfake is a video, photo or audio recording that has been distorted, manipulated or synthetically created using artificial intelligence (AI) techniques to present an individual saying or doing something that they did not say or do. For example, you may unexpectedly receive a video call from someone who looks and sounds exactly like your boss, urgently asking you to wire thousands of dollars to a random account. Although deepfakes have been used for social sharing and entertainment, they are increasingly employed in hoaxes, fraud, and impersonation.

According to a recent study by the biometric identity verification provider iProov, only 0.1% of 2,000 U.S. and U.K. residents could correctly identify deepfake versus real media presented to them, even after being told some of the media would be fake. Participants were also 36% less likely to correctly identify a deepfake video compared to a still image.  

How can you spot a deepfake? Though the technology is constantly improving, certain red flags may help you identify a potential deepfake. Here are a few examples:

  • Unnatural or inconsistent facial expressions: Deepfakes often struggle to replicate subtle human expressions and movements, such as the crinkling of the eyes when a person smiles.
  • Odd eye movements or blinking patterns: Deepfakes may have difficulty with natural blinking or eye movement.
  • Blurring or distortion around the edges of the face: This can be a sign of manipulation as deepfake algorithms struggle with realistic edges.
  • Inconsistent lighting and shadows: Real people are lit by their environment, but deepfakes may not accurately replicate this.
  • Missing or inconsistent features: Features like birthmarks, scars, or makeup can be inconsistent or disappear in deepfakes.
  • No tongue: Some deepfakes have difficulty rendering a tongue when the subject opens their mouth.
  • Lip-synching issues: Pay attention to whether the lip movements match the audio.
  • Audio irregularities: Deepfake audio may sound robotic, have unnatural pauses or inconsistent pronunciation.
  • Background anomalies: Look for distortions, unusual artifacts or inconsistencies in the background.

Now that you know what to look for, what should you do if you think someone is using deepfake technology against you? 

  • Regardless of who you’re talking to (e.g., romantic interest, potential or current employer, even a relative) you should cease contact immediately and only speak through a trusted method.
  • Never provide any personal identifying information or bank account information to someone you’ve met online, even if they’ve shared photos that seem real or engaged in a video call. These can easily be faked.
  • If you believe someone is using deepfake technology on social media to misrepresent facts or give the appearance someone said or did something they didn’t, use the application’s reporting mechanism to prompt a review. 

If you believe you’ve been the victim of a deepfake scam, contact the FBI’s Internet Crimes Complaint Center (ic3.gov) to report your suspicions and any potential losses, and if you believe your MidFirst account may be compromised, contact us at 888.MIDFIRST (888.643.3477) for guidance.

Sources: