Deepfakes only affect video
Deepfake technology encompasses audio cloning, which has been used in documented fraud cases costing millions of dollars. Audio deepfakes have been used to impersonate corporate executives, a U.S. president, and school administrators, causing real-world harm.
What we know
Voice cloning technology uses AI to replicate a person's voice from a small audio sample, often as little as a few seconds of publicly available audio. This technology has moved beyond video manipulation and into audio-only fraud. In 2019, criminals used a deepfake voice of a CEO to instruct a UK subsidiary's CFO to transfer $243,000, believing they were following orders from their parent company's chief executive.
In 2024, attackers used a sophisticated deepfake video conference with multiple fake personas including a synthesized CFO to defraud Arup, an engineering firm, of $25 million. In January 2024, an audio deepfake of President Biden was used in robocalls sent to New Hampshire voters urging Democrats not to vote in the primary. Also in 2024, a high school principal in Baltimore was targeted by an audio deepfake designed to make it appear he had made racist statements, leading to threats against him.
These cases demonstrate that deepfakes are not limited to video manipulation of celebrities or politicians. Audio deepfakes are particularly dangerous because they can be deployed over the phone without any visual component, bypassing visual skepticism, and can be produced much more easily than convincing video deepfakes.
Common claims
- Deepfakes are only a problem for politicians and celebrities in videosFalse. Audio deepfakes have been used in documented corporate fraud cases costing millions.
- You can always detect a deepfake by looking carefullyIncreasingly false. Audio deepfakes require no visual component, and video deepfake detection is an active arms race.
- Deepfake technology is too expensive for most fraudstersFalse. Commercial voice cloning services are widely available and inexpensive or free.