Artificial Intelligence1 min read

I Don't Trust Visual Recordings Anymore. But One Day I Will Again

By · Published by Everything Blog

In short

The shift toward presuming visual content is fabricated gained momentum after an Iranian news agency broadcast disputed missile strike footage that required labor-intensive manual verification. Digital forensics expert Hany Farid eventually confirmed the clip's authenticity by measuring pixel sizes, audio delays, and missile trajectory, but he warned such evaluations cannot match internet scale. As a result, technical leaders are turning to automated verification methods that document an asset's origin at capture. These solutions rely on cryptographically signed metadata and invisible watermarking to restore trust.

Key points

  • Manual forensic evaluations are currently incapable of operating at the massive speed and…: Manual forensic evaluations are currently incapable of operating at the massive speed and scale required by the internet.
  • The proliferation of generative artificial intelligence has placed the primary burden of…: The proliferation of generative artificial intelligence has placed the primary burden of proof on content creators to demonstrate authenticity.
  • Technologies like Google's SynthID aim to protect media integrity by embedding impercepti…: Technologies like Google's SynthID aim to protect media integrity by embedding imperceptible watermarks directly into audio and image pixels.

Sources checked by our analysis pipeline: 1

In March of this year, a flurry of emails and text messages from a half dozen reporters around the world disrupted my quiet Saturday morning. An Iranian news agency had aired a week-old video of what appeared to be a U.S.-made Tomahawk missile hitting a school. News reports claimed that the strike killed more than 175 people, most of them children. The U.S. president and military denied responsibility. In the age of generative AI and all things deepfake, everyone—including me—was skeptical of the video’s authenticity.

I get this type of request a lot. For the past three decades, I’ve been developing mathematical and computational techniques to analyze and detect manipulation in digital media. Over this time, I think it’s fair to say that I, like most people, have approached visual recordings with a “presumption of authenticity”: If a skeptic claims that a recording is fake, the burden of proof is on them.

In today’s world of generative AI, however, I think we need to shift this stance to a “presumption of fake.” That is, the burden is now on the…

Artificial Intelligence