How It Works
The word “deepfake” is a mash-up of the words “deep learning,” which is associated with machine learning and artificial intelligence (AI), and an old-fashioned reference to something being “fake.”
Fraudsters start by taking advantage of an abundance of content available of a specific individual for free from online news outlets and social media, including consumer-generated video and photos. The fraudster manipulates that content into a new image or voice recording with the subject saying or doing something that they never did or that is out of character for them.
The process itself is very technical: Deepfakes are generated by neural networks, using artificial intelligence and machine learning algorithms to imitate real humans. However, an advanced degree is not required to create deepfakes thanks to readily available software that enables people with basic computer skills to become deepfake creators. Quality of the final product may reflect skill level, but may also improve over time due to machine learning, not necessarily human learning.
Who Gets Hurt
Those most often victimized are politicians, celebrities and high-level business executives, due in part to the availability of images, videos and audio recordings of them on the internet. Sometimes the goal of the fraudster is a funny prank that drives social media engagement. But other times the goal is to spread misinformation, ruin reputations or inflict financial harm.
Don’t assume you won’t be impacted by deepfake if you are not a Hollywood or Washington, D.C., socialite. An internet search on deepfakes highlights a wide array of fraud. For example, workers at corporations have been duped into taking inappropriate actions when instructed by a deepfake of their CEO instructing them to do so, sometimes resulting in significant costs and job losses. At a minimum, you may be misinformed about an important topic or person due to deepfakes.
How To Detect Deepfakes
Deepfake images, audio and video can be so well done that it is very difficult to determine a fake. Because deepfakes are synthesized using high tech, detecting them with high tech and algorithms is key. In general, here are a few things to be on the lookout for:
- Poorly synced sound and video, especially with lip movement
- Blurriness where the face meets the neck, hairline or blurry teeth
- Box-like shapes or other cropped effects around the mouth, eyes and neck
- Face discolorations
- Irregular blinking or no blinking
- Asymmetries or inconsistencies in clothing, glasses, jewelry, ears and fingers
- Movements that are unnatural
- Exaggerated, unrealistic or not in-character behavior
- Changes in the background and/or lighting, including monotone or out-of-focus backgrounds
- Lower-quality sections in the same video
Important Legal Disclosures & Information
These articles are for general information purposes only and are not intended to provide legal, tax, accounting or financial advice. PNC urges its customers to do independent research and to consult with security, financial and legal professionals before making any financial decisions. This site may provide reference to internet sites as a convenience to our readers. While PNC endeavors to provide resources that are reputable and safe, we cannot be held responsible for the information, products or services obtained on such sites and will not be liable for any damages arising from your access to such sites. The content, accuracy, opinions expressed and links provided by these resources are not investigated, verified, monitored or endorsed by PNC.
Read a summary of privacy rights for California residents which outlines the types of information we collect, and how and why we use that information.
