Deepfakes and AI
How artificial intelligence is transforming the creation and detection of disinformation.
The AI Disinformation Threat
Generative AI has dramatically lowered the cost of producing convincing fake content. What once required a Hollywood studio can now be done on a laptop:
Deepfake video. AI can generate realistic video of anyone saying anything. In 2022, a deepfake of Ukrainian President Zelensky ordering soldiers to surrender circulated briefly before being debunked. As the technology improves, real-time deepfakes become possible.
Synthetic text. Large language models can generate persuasive articles, social media posts, and comments in any language at enormous scale. A single operator can now produce more unique, human-sounding content than the entire Internet Research Agency.
Cloned audio. Voice cloning can replicate anyone's voice from a few minutes of sample audio. In 2024, a robocall using a cloned voice of President Biden urged New Hampshire voters not to vote in the primary.
Synthetic images. AI-generated photos of fake people, events, and documents can be produced in seconds. Detection tools exist but lag behind generation capabilities.