I have some questions for Mark Zuckerberg
Meta's pitch on AI safety and teen protection doesn't hold up to scrutiny — and the contradictions matter for how we regulate both social media and AI.

Why it matters
Vergecast host Nilay Patel critiques Mark Zuckerberg's recent podcast interview on Meta Glasses surveillance, teen safety settlements, and AI business models, exposing inconsistencies in Meta's stance on regulation, aggregation power, and the gap between stated values and actual product behavior. The piece surfaces uncomfortable questions about market-driven safety claims in AI when the same argument failed for social media.
The key facts
15 to knowMeta Glasses have a recording indicator light, but Zuckerberg frames it as equivalent to phone cameras despite material differences in physical disclosure
Meta has down-ranked videos of retail worker harassment made with Meta Glasses on Instagram, suggesting internal awareness of harms
Zuckerberg claims teen safety research and transparency hurt Meta competitively vs. encrypted competitors like Apple, but internal emails show he questioned research value
Meta estimated $1 trillion in potential liability in recent teen safety case but claims market forces alone will solve AI safety
Zuckerberg describes Muse business model as taking small cuts from businesses in aggregator-style transactions, mirroring Apple's 30% App Store fees Meta has publicly opposed
Louisiana Hyperion data center secured through backroom negotiations and property tax exemptions, not public benefit disclosure
Zuckerberg argues market will kill harmful AI products ('won't be able to happen'), same argument rejected for social media harms
Meta settled teen safety litigation; Zuckerberg cited trillion-dollar liability estimate in internal documents
Zuckerberg argues Meta Glasses passive recording should be treated like smartphones, but analyst counters the physical action difference is material
Meta has down-ranked harassment videos of retail workers and women recorded on Muse to reduce incentive for similar content
Zuckerberg proposed hour-per-day teen usage limits but framed it as a prisoner's dilemma requiring unilateral government regulation across all platforms
On AI, Zuckerberg posted that trust and alignment are differentiators and the market will enforce safety; Meta delayed Muse by months for security
Zuckerberg frames Muse business model as aggregation: small take-rate from businesses, not consumers, analogous to Apple's App Store model
Meta's Louisiana Hyperion data center deal involved tax exemptions and backroom negotiations not publicly disclosed upfront
Analyst notes Instagram is beloved by consumers despite harms; unclear whether Muse or other AI products will achieve sufficient consumer enthusiasm to overcome skepticism about data-center environmental/social costs
The story so far
Earlier coverage of this storyline
- AI safety conversations have gotten unbelievableTechCrunch AI
- Nvidia CEO Jensen Huang emerges as Trump's top ally in AI safety debateCNBC Technology
- Introducing MentalHealthBenchOpenAI Blog
- How to think about the end of the worldFinancial Times Technology
- This story
Go to the source
The Verge AItheverge.com
Publisher excerpt: It’s a big week for Meta. The company just kicked off its big Connect conference on Wednesday, and the new Muse AI agent appears to be an early hit. I’ve been using it — it is surprisingly good. CEO Mark Zuckerberg has obviously been out there taking a victory lap and participating in a lot of…