Urgent.News

What's breaking now, across thousands of outlets.

Tech

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

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…

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

In March of this year, a wave of emails and text messages swept through newsrooms worldwide. An Iranian news agency had broadcast a video from a week prior that purportedly showed a U.S.-made Tomahawk missile striking a school, causing over 175 casualties, most of them children. The U.S. government denied involvement. In an era of deepfakes and generative AI, skepticism about the video's authenticity was rampant.

The author of this piece, a researcher specializing in detecting digital manipulation, has been working in this field for three decades. Initially, like most people, they assumed videos to be authentic. However, in today's landscape, they believe the opposite - that the burden lies on the creator to prove a video's authenticity.

This shift in perspective reflects reality and suggests a future where online media can once again be trusted. The researcher's initial analysis of the missile video suggested that the missile trajectory was physically implausible if artificially added. They also questioned the missile's pixel size, estimating it to correspond to a 6-meter Tomahawk at approximately 100 meters distance.

However, after further analysis, they concluded the video was authentic. This shift brought the researcher full circle, moving from a presumption of authenticity to a presumption of fake. To counter the growing issue of fake videos, provenance signals like content credentials and watermarks can be used. These signals, such as the C2PA standard and Google's SynthID, embed information about the content's origin and any edits into the media.

While these mechanisms may not be completely foolproof, they empower legitimate creators and help distinguish genuine content from manipulated versions.

Written by urgent.news from IEEE Spectrum's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at spectrum.ieee.org →

More in Tech

P5 - Permissions in OKE: the node has identity, and all its pods inherit it

"Kubernetes on OCI at the lowest cost" series — part 5. In part 4 we set up the cluster and node pools. Here's how what's running inside gets permissions on OCI.

  • Nodes in OKE cluster receive permissions via their instantaneous identity
  • Two types of identity in OCI: users with own keys and principal instance
  • Each pod inherits permissions based on applied policy upon creation

More from Friday 20 November →