The 404 Media Podcast – Episode Summary

Episode Title: A Massive Breach Reveals the Truth Behind 'Secret Desires AI'
Date: November 26, 2025
Hosts: Joseph (host), Sam Cole, Jason Kebler, Emanuel (mentioned)
Main Theme:
This episode centers on an investigative story by Sam Cole about a major data breach affecting the erotic AI chatbot platform "SecretDesires.AI." The team unpacks how the breach exposed the surreptitious conversion of women’s yearbook and social media photos into non-consensual AI-generated pornography, revealing a troubling picture of abuse, privacy failure, and industry incentives.

EPISODE OVERVIEW

The episode’s main focus is Sam Cole’s report on SecretDesires.AI, where a leaked database provided a window into how users are exploiting AI tools for generating non-consensual sexual imagery, including using ordinary people’s yearbook photos. The hosts dissect the technical details of the breach, its implications for privacy, the disturbing content discovered, and the broader societal and legal questions raised. Later, Jason Kebler discusses viral revelations about bot and spam accounts on X (formerly Twitter) and the international "side hustle" economy targeting U.S. social media users.

KEY DISCUSSION POINTS & INSIGHTS

1. What is "SecretDesires.AI"? (04:17 – 09:16)

• Overview: SecretDesires.AI is an erotic AI chat and image generation platform similar to other sex-driven chatbot/image generator products.
• Functionality: Users customize chatbots’ appearance and personality, including generating images—99% of which are erotic.
• Image Generation: Includes explicit categories by ethnicity, age, style (anime/realistic), and a feature for user-uploaded images (“face swap”), though this was recently removed.
• Demographics: While the homepage mostly features hypersexualized anime women, the user base includes both men and women. Sam notes:

  “There’s a big demographic of women using erotic chatbots and role play...it’s a lot of the same draw as a romantasy novel except you’re interacting with the character.” (05:23 – 06:33)

• Shift from Deepfakes: Modern tools now synthesize whole new scenarios and bodies, not just face swaps.

2. The Massive Data Breach & How It Was Discovered (10:05 – 11:58)

• Source: Sam received a tip via email about a public, unprotected Microsoft Azure Blob Storage container.
• Scale: Contained ~1.8 million images (including duplicates), both source (“input”) photos and resultant AI-generated (“output”) images.
• Types of Input Images:
  • Old MySpace-era pictures
  • Flip phone and low-res photos
  • Screenshots of influencers, celebrities, politicians
  • Yearbook and graduation photos
• Quote:

  “It was like completely random people...a lot of them old images...deep fried, low-resolution, like MySpace picture.” (12:55 – 14:22)

• Notable Case: Sam found a local politician’s town hall photo uploaded then used to generate explicit imagery.

3. How the Platform Was Abused (14:50 – 17:47)

• Face Swap Tool: Enabled users to upload someone’s photo (e.g., strangers, acquaintances, influencers) and generate custom pornography with that likeness.
• Non-Public Use: Even if the generated content isn't posted on the platform, it can be freely shared elsewhere.
• Quote:

  “Someone could upload a picture of this girl at the gym… [generate the content]… and just because you can’t post it on the public feed doesn’t mean you can’t share it other places.” (14:50 – 17:47)

4. The Breach’s Revelations: Ordinary People Targeted (17:47 – 19:52)

• Snapshot of Hidden Activity: The database gives rare, concrete insight into the abuse of generative AI for private, non-consensual pornography.
• Yearbook Photos: The story’s headline used “yearbook” to highlight the recontextualization of innocent photos.
• Quote:

  “Literally like pictures you would see [in a] yearbook. Someone graduating, class of 2010, holding the little cap, LifeTouch background…” (19:20 – 19:41)

5. How Explicit Was the Content? (20:22 – 23:53)

• Very Extreme Content:
  • Scenarios “not humanly possible”
  • Included minors and obviously underage likenesses
  • Violent or extremely hardcore AI-generated scenes
• Legal Risks: Generation of AI child sexual abuse material (CSAM) is a prosecutable crime in many jurisdictions.
• Quote (Jason on editing the piece):

  “What I saw was extreme...almost all of it. This is not like a standard nudify app...it is like for hardcore porn primarily.” (23:20 – 23:53)

6. Emotional Responses & Takeaways (23:53 – 29:34)

• Jason:

  “It’s one of those moments that makes you lose a little bit of faith in humanity...people are doing this. That’s not good. And we’re not really doing anything to stop it.” (25:50 – 26:16)

• Sam:
  • Scale and ordinariness: the most disturbing revelation was sheer prevalence and the targeting of regular people.
  • The face swap tool’s removal suggests the company had second thoughts—perhaps due to legal risks or new legislation.
• Company Response:
  • Rapidly closed exposed data buckets after being alerted by Sam.
  • Did not issue a public statement despite aggressive advertising of fantasy/face swap capabilities.

7. Wider Tech & Societal Context (29:34 – 34:42)

• Broader Implications:
  • Journalists repeatedly find “exposed bucket” security disasters—a chronic industry problem.
  • The story shows tech’s role in scaling privacy violations and possible criminal abuse at speeds and scales never before seen.

8. The X Geotag Scandal & The Side Hustle Economy (34:42 – 53:34)

X’s Geotag Reveal (35:00 – 41:21)

• New Feature: X (Twitter) adds country-of-origin visibility for accounts, revealing many viral “American” meme/political accounts are run from overseas.
• Examples: “Ivanka News” (Nigeria), “Red Pill Nurse” (Eastern Europe), “MAGAnadine” (Morocco), “Native American Soul” (Bangladesh).
• Immediate Fallout:
  • Online freakout as users realize trusted accounts are not what they seemed.
  • Concerns over disinformation, trolling, and commercial manipulation.
• Jason’s Take:

  “This is not a new feature for social media...and I would argue, it hasn’t really done anything to better our online discourse or make people understand that not everyone is telling the truth on the internet.” (36:54 – 38:46)

The Global Hustle to Monetize U.S. Social Media (42:01 – 53:34)

• Root Cause:
  • U.S. social media is the world’s most valuable ad market due to high CPMs and lax privacy laws.
  • Content creators globally seek to game algorithms, target U.S. audiences, and maximize ad/creator payout.
• Ecosystem:
  • YouTube/FB/Telegram/TikTok are flooded with tutorial videos on making “USA channels” using AI voice, video, and text tools.
  • Strategies include copying trending American topics, using generative AI for content, and recycling news/history/cultural themes.
  • Quote:

    “There are dozens and dozens of videos about how to make USA content...CNN is a popular American news source. Go to CNN.com, copy-paste their articles, put them into ChatGPT, change it, have it write you a script.” (45:14 – 47:23)

• Spam & Disinfo:
  • The viral content often feels “culturally off”—AI or translation errors abound.
  • Most creators are motivated by higher payouts for U.S. viewers, not ideology or politics.
• Money Quote:

  “You’re making between five and seven times as much...if you’re able to reach an American audience. Why would you cater to a smaller audience when you can spam American topics?” (52:26 – 53:34)

• Implications: Monetization tools and lack of platform oversight have created a global spam gold rush centered on the U.S., incentivizing misinformation and low-quality content.

NOTABLE QUOTES & MEMORABLE MOMENTS

• Sam Cole [14:26]:
  “The name of the container is how we figured that out, because they’re called faceswap...people would then upload the images… [and] SecretDesires was using its generative AI algorithm to turn those into whatever fantasy prompting the user described.”
• Jason Kebler [23:20]:
  “The stuff that I saw when editing this piece was extreme. Like, extreme almost all of it...this is not like a standard nudify app...it is for hardcore porn primarily…”
• Sam Cole [26:16]:
  “It’s just something that we know people have been doing...fantasy is fantasy, porn is fantasy...But with the face swapping stuff...it was very telling that they took it down...people really want this feature.”
• Jason Kebler [25:50]:
  “It’s one of those moments that makes you lose a little bit of faith in humanity...people are doing this. That’s not good. And we’re not really doing anything to stop it.”
• Jason Kebler [52:26]:
  “You’re making between five and seven times as much...if you’re able to reach an American audience...so why would you cater to a smaller audience...when it takes the same amount of effort to spam something about the NBA or American politics?”

IMPORTANT TIMESTAMPS

| Segment | Timestamp | |--------------------------------------------------------|:-------------:| | Intro, subscriber note, survey talk | 00:00 – 03:35 | | What is SecretDesires.AI and the AI porn ecosystem | 04:17 – 09:16 | | Technical breach, exposed data, investigation | 10:05 – 11:58 | | User–generated abuse: face swap and real people | 14:50 – 17:47 | | Yearbook photos and ordinary people abused | 19:20 – 19:52 | | How explicit and illegal was the AI content | 20:22 – 23:53 | | Editorial reactions, company “response” | 23:53 – 29:34 | | X’s geotag feature & user backlash | 35:00 – 41:21 | | International engagement farming & AI slop economy | 42:01 – 53:34 |

SUMMARY

This episode exposes the dark reality and immense scale of non-consensual AI-generated pornography, as reckless data storage and generative AI converge. Through a rare, raw journalistic window (thanks to a data leak), the team connects the dots between modern sex tech, privacy failures, and the global social media economy—where the pursuit of ad dollars often trumps any ethical concern. The episode blends detailed investigative reporting, technical analysis, and cultural critique with a conversational, irreverent tone that underlines the surreal stakes and human cost of these new digital frontiers.