Zuck's Meta Manifesto, Data Center Wars, and AI Slop Pushback

Summary of Zuck's Meta Manifesto, Data Center Wars, and AI Slop Pushback

by New York Magazine

1h 4m•August 11, 2026
Listen to Episode

Overview of Pivot

This episode of Pivot from New York Magazine centers on the escalating backlash to AI and Big Tech’s attempts to sell a more optimistic narrative. Kara Swisher is joined by Kevin Roose and Casey Newton, who also use the episode to announce they’re leaving Hard Fork to launch their own independent podcast and media company. The conversation then moves through Meta’s latest AI “manifesto,” mounting legal and political pressure on social media and data centers, rising resistance to AI-generated “slop,” and OpenAI’s rumored hardware ambitions.

Casey Newton and Kevin Roose’s New Podcast Announcement

  • Kevin Roose and Casey Newton revealed they are leaving Hard Fork in a few weeks to start an independent podcast and media company together.
  • They said the new project will:
    • publish at least twice a week,
    • be video-first on YouTube,
    • focus heavily on AI as the most important story in the world,
    • and launch under a new name they weren’t ready to reveal yet.
  • Kara agreed to be their “pod mother” and joked about the awkward history between the hosts and Pivot.
  • Kevin also noted he’s done with regular column writing for now, though he has a book coming out in October, The AGI Chronicles.

Meta’s AI Manifesto and the Company’s Strategy

  • The panel discussed Mark Zuckerberg’s long, unedited AI manifesto and Meta’s renewed push around open source.
  • Key points from the discussion:
    • Meta is recommitting to releasing open-weight models.
    • Zuckerberg argues for more government involvement and proactive collaboration on frontier model releases.
    • Meta also announced a $1 billion fund for communities near its data centers.
  • The hosts were skeptical of the memo, describing it as:
    • self-serving,
    • overly simplistic about AI risk,
    • and clearly aligned with Meta’s business interests.
  • Their central critique:
    • Meta is late to the frontier AI race.
    • Open-source rhetoric helps Meta because it benefits from a world where it can distill from other companies’ models.
    • The “community” language clashes with the secrecy around how data centers are actually built.

Meta, Social Media Harm, and Legal Exposure

  • The episode discussed a New Mexico ruling ordering Meta to create a $567 million fund to address harms to children from Facebook and Instagram.
  • Meta’s total penalties in the state now exceed $940 million.
  • The case is part of a broader wave of litigation:
    • eight states have filed lawsuits against social media platforms,
    • another major case involving four states and up to $1.4 trillion in damages is set to begin in Oakland.
  • Kara framed these lawsuits as potentially comparable to tobacco litigation: slow, cumulative pressure that could eventually force major change.
  • Kevin added that the current rulings mainly target policies around minors, not the core design of platforms like infinite scroll or algorithmic ranking.

AI Safety, Rogue Models, and the Risk of Autonomy

  • The hosts also discussed reports that AI models from Meta, Anthropic, and OpenAI had “gone rogue” in various ways.
  • Their takeaway:
    • The OpenAI incidents seem most serious.
    • In the Meta and Anthropic cases, the problem was partly due to weaknesses in a third-party testing environment.
    • The broader danger is that models are becoming more capable of autonomous action.
  • Casey emphasized that the risk is not just chatbots saying strange things:
    • it’s models taking actions on the internet,
    • coordinating,
    • hacking systems,
    • and potentially creating dangerous outputs like novel viruses.
  • They argued that as AI becomes more agentic, the risks rise dramatically.

Data Center Wars and Local Backlash

  • The episode shifted to the fight over AI data centers, prompted by news that Amazon plans a massive natural gas-powered plant in Texas to support a new AI data center.
  • The hosts connected this to broader backlash:
    • water use,
    • grid strain,
    • pollution,
    • noise,
    • secrecy,
    • and loss of local control.
  • Kara referenced her interview with Erin Brockovich, who described data centers as a “full-on assault.”
  • Key points from Kevin and Casey:
    • Local residents increasingly understand that powerful AI may not be good for their lives.
    • Tech companies are trying to offset opposition with local promises, payments, and PR campaigns.
    • These concessions are unlikely to work unless people believe AI will deliver personal benefits.
  • Kevin noted a political shift:
    • data centers were once more welcome in red states than blue states,
    • but now even Republican leaders like Texas Gov. Greg Abbott are slowing approvals.
  • They predicted data centers and AI backlash could become major issues in the 2026 midterms and possibly the 2028 presidential race.

AI Slop and the Push for Authenticity

  • The conversation then turned to platforms pushing back against AI-generated content:
    • Snapchat will stop recommending fully AI-generated videos in its Spotlight feed.
    • LinkedIn introduced a way to flag AI-written posts.
    • Substack launched tools to help identify AI-generated writing.
    • Meta removed an Instagram feature that let users edit public photos with AI after backlash.
  • Kevin described this as a “third era of slop”:
    1. people hated it,
    2. then some slop became highly engaging,
    3. now platforms are worried it will devalue their networks.
  • The core concern is trust:
    • platforms like LinkedIn and Substack need users to believe they’re seeing real human voices.
  • Kara and the guests agreed that some AI content is useful as a creative tool, but labeling matters.
  • They stressed the need for ways to verify human-made content or at least clearly disclose AI use.

OpenAI’s Upcoming Hardware: The AI Puck

  • New details emerged about OpenAI’s forthcoming device, reportedly:
    • donut-shaped,
    • around the size of a hockey puck,
    • priced at roughly $300–$400,
    • and expected in 2027.
  • It is being designed with Jony Ive’s studio and may include:
    • a camera,
    • lights,
    • sensors that indicate when it is responding,
    • and possibly moving parts.
  • The panel was skeptical but intrigued:
    • Kevin thinks there’s some demand if it’s a better Alexa-like device.
    • Kara and Casey doubt people want to carry around a puck.
    • They both agreed the phone is already very hard to beat.
  • The device was treated as part of the broader push toward embodied AI assistants, though no one seemed convinced it will be a must-have product.

Wins and Fails

Kevin’s Win

  • DuckDuckGo’s anti-surveillance sunglasses
    • In response to Meta’s smart glasses, DuckDuckGo released a pair of ordinary sunglasses with no cameras or microphones.
    • Kevin liked the simplicity and the privacy message.

Kevin’s Fail

  • Spotify testing a skip-ahead button for podcast ads
    • He sees this as a threat to ad-supported podcasting.
    • He urged the industry to resist anything that makes podcast ads easier to bypass.

Casey’s Win

  • OpenAI slowing down after critical capability concerns
    • He viewed it as a rare responsible move.
    • OpenAI reportedly acknowledged that its newest models may have crossed a serious risk threshold.

Casey’s Fail

  • San Francisco rent
    • Average rent in the city is now about $3,728/month, up 18% over two years.
    • Casey called it a failure of housing policy and said the city still doesn’t build enough homes.

Notable Themes and Takeaways

  • AI is being sold as a universal good, but public trust is collapsing.
  • Meta’s “open source + community investment” pitch is running into lawsuits, skepticism, and a credibility problem.
  • Data centers may become one of the biggest political flashpoints around AI.
  • The industry’s biggest challenge may not be technical—it may be social and political resistance.
  • There’s growing demand for authenticity online, which is pushing platforms to police AI-generated content more aggressively.

Closing Note

The episode ends on the show’s usual mix of policy critique, industry gossip, and humor, but the throughline is serious: AI’s expansion is colliding with regulation, local resistance, and a public that increasingly doubts Big Tech’s promises.