Skip to content

Catfishing, Airbnb fraud, and other Deep Dreams about catching computer fakes

Endless Thread

Hosted by Amory Sivertson and Ben Brock Johnson

27m
Transcript

Original podcast

Where to listen

Summary

Written by a model from the transcript; not yet reviewed.

This episode explores how artificial intelligence has evolved from early computer vision projects into modern deepfakes and image generators, and examines how internet communities are fighting back against AI-driven scams and misinformation. The host, Ben Brock Johnson, walks co-host Amory Sivertson through the history of Google's 2014 Deep Dream project, explaining how teaching computers to categorize millions of images paved the way for today's hyper-realistic AI art and video. They discuss the massive online subreddit Is This AI?, where millions of users crowdsource the detection of digital fakes, ranging from catfishing scams and fake Airbnb damage claims to political disinformation. Finally, the hosts examine watermarking technologies like Google's Synth ID and proposed federal legislation as potential tools for verifying digital media provenance.

  • Check suspicious images using detection tools and watermarking technologies like Google's Synth ID to verify if they were generated by AI. 16:45
  • Browse online communities such as the Is This AI subreddit to see crowdsourced detective work on potential digital scams and fakes. 17:27
  • Contact your federal representatives to support pending legislation regarding artificial intelligence labeling and deepfake regulations. 28:36

Chapters

  1. 0:00
    Podcast Intro and Sponsorships

    Hosts introduce the episode, sponsors, and set the stage for the discussion.

  2. 1:23
    Hosts Introduce AI Image Detection Topic

    Ben and Amory discuss their feelings about AI-generated images and the challenges of spotting them.

  3. 2:15
    Deep Dream Origins and Computer Vision Basics

    The hosts explain the history of Deep Dream and foundational concepts of computer vision.

  4. 11:55
    Challenges and Future of AI Detection

    Hosts reflect on the limitations of current detectors, the need for legislation, and the role of human judgment.

  5. 14:05
    The Is This AI? Subreddit and Real-World Scams

    Discussion of how the subreddit helps users spot AI-generated images in scams such as Airbnb fraud and catfishing.

  6. 17:04
    Evolution of AI Image Generation and Detection

    They trace progress from early AI art to modern deep fakes and the rise of detection tools like SynthID.

  7. 21:01
    SynthID: How AI Watermarking Works

    Explains how SynthID detects AI-generated images by looking for hidden patterns rather than visual cues.

  8. 29:16
    Closing Remarks and Call to Action

    Final thoughts, resources, and encouragement for listeners to engage with AI labeling legislation.

Highlights

  • 1:31

    Evolution of AI Image Flawsinsight

    The guest notes that early AI image artifacts like extra fingers are disappearing, making generated images much harder to spot initially.

  • 4:31

    Origins in Deep Dreamstory

    The host explains that Google's 2014 Deep Dream computer vision project used convolutional neural networks to teach machines how to see and categorize images.

  • 16:45

    Airbnb Fraud Detectionstory

    A Reddit post on the Is This AI subreddit exposes an Airbnb host trying to charge a guest for damages using an AI-generated image.

  • 17:27

    Rapid Growth of Detection Subredditdata

    The Is This AI subreddit exploded in popularity after launching in May 2023, drawing three and a half million weekly visitors.

  • 20:11

    Unreliability of Standard AI Detectorsinsight

    Reddit contributors point out that traditional AI image detectors are janky, unreliable, and often misidentify genuine older photos as artificial.

  • 21:48

    How Synth ID Watermarks Workstrategy

    Synth ID embeds a hidden pattern directly into Google-generated images, allowing tools to check for specific creation tools rather than visual artifacts.

  • 28:36

    Proposed Federal AI Legislationdata

    Congress is considering several bills, including the AI Labeling Act of 2026 and the REAL Act, to address deepfaked and edited media.

Quotes

  • “Deep Dream was a computer vision project from way back in 2014.”
    Ben Brock Johnson · 2:41
  • “You don't just need to rely on big tech companies to build tools that may soon become obsolescent.”
    Ben Brock Johnson · 28:28
  • State Farm 0:00–0:59
  • State Farm 13:21–13:46

About this episode

What do you do when the uncanny valley of detecting deep fakes and other AI-generated imagery becomes harder and harder? As the flood of slop rises towards our collective nostrils, a combination of crowd wisdom and tools built by the tech companies driving AI seem to be the current answer. Today on Endless Thread, Ben takes Amory on a brief journey through the recent history of computer vision, and finds some optimism in a few tools that are becoming more reliable in helping us gut-check a host of new AI-powered scamming. Is this AI? Listen on to find out! Show notes: Computer Vision http://Computer vision (Wikipedia) Deep Dream https://en.wikipedia.org/wiki/DeepDream# (Wikipedia) Air BnB toilet https://www.reddit.com/r/isthisAI/comments/1wt6sb0/photo_of_a_toilet_airbnb_sent_in_justification/#lightbox (Reddit) R/IsThisAI SynthID explainer https://www.reddit.com/r/isthisAI/comments/1uin017/psa_synthid_what_is_it_how_do_i_check_and_what/ (Reddit) This content was originally created for audio. An auto-generated transcript is available on Apple Podcasts and Spotify. Heads up that some elements (i.e. music, sound effects, tone) are harder to translate to text.

Transcript

0:00
27:44

4,594 words · 8 speakers

Speaker names marked are inferred from speaking time and the show’s known hosts — they are not voice-verified.

Topics

  • Google
  • Synth ID
  • Is Business Broken
  • OpenAI
  • Is This AI?
  • AI images
  • Deep Dream
  • Gemini
  • WBUR
  • Marotra Institute

Page updated · transcript and summary are machine-generated