The 404 Media Podcast

Episode: The Disappearing DOGE Depositions

Date: March 18, 2026
Hosts: Joseph Mayberg, Emmanuel Mayberg, Jason Kebler

Episode Overview

This week’s 404 Media Podcast dives into two major stories:

1. The DOGE Depositions: Six hours of revealing legal testimony from key members behind the controversial government grant-cutting campaign—initially made public, quickly going viral, and then effectively disappearing after a judge’s order.
2. AI’s African Backbone: On-the-ground reporting from Kenya where the workers who labor behind AI systems are organizing for better labor conditions, challenging the global tech industry’s exploitative frameworks.

Throughout, the hosts bring trademark 404 Media clarity and skepticism, pulling stories from the digital shadows into the light.

Segment 1: The Disappearing DOGE Depositions

(Starting ~03:25)

1. Background and Context

• Story Recap: Joseph watched and reported on six hours of deposition footage from DOGE members, Justin Fox & Nate Kavanagh.
  • DOGE orchestrated massive cuts to U.S. government grants via the NEH.
  • Lawsuits have followed, filed by major academic associations (Modern Language Association, American Council of Learned Societies, American Historical Association).

2. Key Moments from the Depositions

• Availability and Rarity: The depositions were public, uploaded to YouTube—"somewhat rare" (05:00).

• Nature of the Depositions: Focused on in-depth questioning, unlike previous public appearances (05:21).

• DEI Definition Evasion:

  • Fox unable or unwilling to define “DEI” (Diversity, Equity, and Inclusion), repeatedly citing only the Executive Order as his reference.
  • Key Exchange:
    • Lawyer: “How do you interpret DEI?”
      Joseph (as Fox): “There was the EO [Executive Order] explicitly laid out the details. I don’t remember it off the top of my head.” (07:28)
    • Lawyer: “So can you…”
      Joseph (as Fox): “I don’t remember what was in the...”
  • Joseph’s summary: “He’s incredibly evasive, obviously.” (06:14)

• Specific Examples of Questionable Justifications:

  • Grant for a documentary about Black civil rights was cut because, according to Fox, “it wasn’t for the benefit of humankind.” He later walked back this statement after it was read to him live in the deposition. (08:00)
  • Grant for a Holocaust documentary featuring women survivors was cut due to “the gender based story—that’s inherently discriminatory to focus on this specific group.” (08:34)

• Use of ChatGPT in Review Process:

  • Fox: “...when they used ChatGPT, they were searching for, you know, ‘black,’ ‘homosexual,’ ‘LGBTQ,’ but they didn’t search for ‘white’ or ‘Caucasian.’ And he does acknowledge that.” (11:34)

• Evasiveness and Coaching:

  • The DOGE members’ responses were clearly coached by Department of Justice lawyers present during the depositions. (10:28)

3. Public and Legal Response

• Virality: Clips of the deposition went viral, especially after 404 Media’s coverage and video highlights. (12:48)
• Notable Quote:
  Emmanuel: “It’s like, if there’s any initiative for women, if there’s any documentary that focuses on women, or certain minorities, it’s like, that is DEI and… it’s impossible to do anything.” (10:28)
• Right-Wing Co-optation: Right-wing X (Twitter) accounts began reposting 404 Media’s clips, “That’s the 404 media font.” (13:32)

4. Judicial Intervention and Takedown

• Government Argument:
  • DOJ requested videos be removed, citing “potential harassment and reputational harm” to Fox (13:32).
  • Cited 404 Media's article directly.
  • Alleged Fox faced death threats, though the threats themselves were not public.
• Takedown Ordered: Judge ordered all videos removed late Friday night.
  Joseph: “...to a judge saying, you must remove those from YouTube.” (16:32)
• Memorable Reaction:
  • Jason: “Can we discuss how crazy this is? …these are government employees, in a highly publicized case of great public interest… and we are protecting them because people are being mean to them online.” (16:32)
  • Emmanuel: “...public servants, right? This is about stuff they did for the government...taxpayer funded activities.” (17:43)

5. The Streisand Effect: Video Resurrection

• Backup Surge:
  • Videos quickly appeared on the Internet Archive (Saturday) and have been torrented—now “censorship proof” thanks to decentralized sharing.
  • Jason: “Torrents are undefeated in that way. Like, it will live forever somewhere.” (21:31)
  • Joseph: “...other people had already archived them.” (22:54)
• Memorable Comment:
  • Jason: “It’s the Streisand effect…there’s an initial spike in interest and people kind of like finding something when the government deletes it… But I think that in this case, like, because it’s torrented now, it’s censorship proof.” (21:31)

Segment 2: AI’s African Backbone—The Hidden Human Labor in AI

(Starting ~28:28)

1. On-the-Ground Reporting from Kenya

• Jason’s Trip: Attended a conference, conducted field reporting on data labelers—the often-invisible workforce making AI possible.
• Who Are Data Labelers?
  • Workers tasked with identifying, annotating, and categorizing all sorts of digital data—needed for everything from facial recognition to “training” AI chatbots.
  • Many work for major subcontracting firms like Sama and Appen, often for meager pay and with little support.

2. Notable Stories and Conditions

• Portrait of a Data Labeler:
  • Michael Jeffrey Asia: Former data labeler, now activist.
    • Spent days categorizing porn scenes, then shifted to working as the “AI” behind sex companionship bots (masquerading as various personas for Western users).
    • Jason paraphrasing Michael’s experience:
      “He was paid very little… quite traumatizing because I felt pulled between all these different personas… After over a year… ‘Fuck this. This is terrible. I hate this. We need to fight for better rights.’” (34:55-36:59)
    • Experienced PTSD, marital troubles, insomnia due to trauma and exhaustion.
• Gig Work Ubiquity:
  • Data labeling is as common in Kenya as delivery driving or rideshare in the West.
  • Joseph: “Basically ingrained… maybe the culture is the wrong word, but like data labor in the economy for sure.” (35:54)

3. Organizing and Pushback

• Data Labelers Association:
  • Formed by Michael and colleagues to fight for improved wages, conditions, and legal protections.
  • Jason: “They’re just signing people up… it doesn’t have to be this way.” (37:02)
• Legal Efforts:
  • Lawsuits ongoing in Kenya targeting companies (Sama, Meta, OpenAI) for lack of mental health support, poor pay, and other labor law violations.
  • Jason (quoting lawyer Mercy Mutemi):
    “We have laws that should protect against this. It’s just a matter of getting them enforced.” (37:02-40:00)
• Corporate Threats:
  • Big tech threatens exit if forced to improve conditions—“If you regulate us, we’re just going to leave the country.” (40:00)
• Spectrum of Work:
  • Sensitive roles extend to translation, reviewing audio for tech firms, and even working on U.S. military projects—often without the worker’s knowledge. (40:00-41:56)

4. Cultural and Linguistic Ironies

• AI Language Loop:
  • Many Kenyans accused of “sounding like ChatGPT” in emails/LinkedIn posts—but in reality, AI is writing like them, as Kenyan workers trained much of the AI’s output. (42:04)
  • Jason paraphrasing a Kenyan writer:
    “I don’t write like ChatGPT. ChatGPT writes like me… now when we write in the way we were taught in school, we’re getting accused of using AI.” (42:04)
• Quote of the Episode:
  • Michael (via Jason): “AI is African Intelligence.” (44:58)
  • Jason elaborating:
    “AI is not magic. It’s just zillions of human hours that go into…managing the outputs and tweaking the outputs… The people doing that…a lot of them are African. This is our labor. We’re getting paid $200 a month…and OpenAI is worth a trillion dollars.” (44:58)

Notable Quotes & Timestamps

• On Government Evasiveness (DEI Definition):
  • “He was unable or unwilling to define DEI…incredibly evasive, obviously.” – Joseph (06:14)
• On How These Cuts Happened:
  • “...if it included the name of any minority or DEI or gender, it just got removed.” – Emmanuel (12:07)
• On Judicial Takedowns:
  • “This is not supposed to happen, I don’t think.” – Jason (17:59)
• On The Streisand Effect:
  • “Torrents are undefeated…It will live forever somewhere.” – Jason (21:31)
• On AI Labor and Exploitation:
  • “He was paid very little. It was quite traumatizing…this ruined my life in many ways.” – Jason on Michael Jeffrey Asia (34:55)
• On Kenyan Labor’s Global Impact:
  • “AI is African Intelligence. Everyone knows this. They should know this.” – Jason (44:58)

Key Timestamps

| Timestamp | Segment/Topic | |-----------|------------------------------------------------------------------------------------------| | 03:25 | Introduction to DOGE depositions story | | 05:21 | DOGE depositions—nature and public access | | 07:19 | Evasive DEI definition exchange (detailed Q&A sample) | | 08:34 | Specific grant cuts (Black civil rights, Holocaust documentary) | | 11:34 | Use of ChatGPT in discrimination | | 13:32 | Virality and right-wing appropriation of video clips | | 13:32-16:32| DOJ/judicial takedown request and response | | 17:43 | Discussion of public servant status and free speech implications | | 19:11 | Backup and preservation on Internet Archive/torrents | | 21:31 | Streisand Effect reflections | | 28:28 | Transition to Jason’s reporting in Kenya, introduction to AI workers story | | 29:09 | What is data labeling? | | 32:36 | Portrait of Michael Jeffrey Asia, traumatic labor (sexbot/user Persona work) | | 34:55 | The toll of the work: PTSD, family troubles | | 36:59 | Birth of Data Labelers Association — organizing pushback | | 37:02 | Legal efforts and corporate resistance | | 42:04 | Linguistic ironies: “I don’t write like ChatGPT. ChatGPT writes like me.” | | 44:58 | “AI is African Intelligence” — thesis and call for recognition |

Analysis and Takeaways

• Transparency vs. Censorship: The DOGE depositions controversy underscores tensions between legal privacy, transparency, and the viral nature of digital content—especially when public servants’ actions are involved.
• Invisible Labor in Modern Tech: The Kenyan data labelers’ story reveals how “AI” is reliant on and shaped by human labor, often underpaid and hidden, raising urgent questions about compensation, trauma, and recognition.
• Global Digital Power Dynamics: Both stories expose how decisions (whether about government grants or tech industry automation) made by a small group have far-reaching, sometimes devastating consequences—often shielded from scrutiny until journalists surface them.

Final Thoughts

The episode showcases 404 Media’s commitment to uncovering the seldom-seen machinery behind both government policy and cutting-edge technology, reminding listeners that what disappears from platforms often continues to live on—and that the real “intelligence” in AI may come from those farthest from the spotlight.