Ben Thompson outlined a series of questions for Mark Zuckerberg following the Meta Connect conference, focusing on the company's new Muse AI agent. The Decoder host noted that Meta declined an interview request, prompting him to analyze remarks Zuckerberg made on Alex Heath's Sources podcast. Thompson argued that Zuckerberg's comments reveal significant gaps in the company's approach to privacy, regulation, and market dynamics.
A primary point of contention involves Meta Glasses. Zuckerberg defended the device by comparing it to smartphones, arguing that the physical action of retrieving a phone to record serves as a social cue absent from passive surveillance via eyewear. Thompson countered that this comparison ignores the distinct social norms required for constant recording. He cited internal Meta actions, specifically the down-ranking of videos showing harassment of retail workers and women on Instagram, as evidence that the company acknowledges harmful use cases. Thompson questioned whether Meta believes content moderation alone can resolve issues stemming from the device's design, noting that users define product behavior regardless of initial communication about recording lights.
The discussion extended to teen safety and the recent multi-billion dollar settlement. Zuckerberg suggested that high usage rates indicate user satisfaction and that safety features have been developed over time. Thompson challenged this logic, noting that lawsuits allege product addiction rather than mere popularity. He highlighted a feedback loop where Meta faces legal penalties for transparency about platform harms, while competitors like Apple avoid scrutiny by not publishing similar data due to encryption. Thompson argued that Zuckerberg advocates for government regulation in social media to solve a prisoner's dilemma regarding teen usage limits across platforms, yet opposes similar regulatory frameworks for AI safety.
This contradiction in regulatory stance formed the core of Thompson's critique. In AI, Zuckerberg has argued that market forces and competitive liability will drive safety and alignment, citing Meta's delay of Muse for security improvements as proof. However, in social media, he calls for unilateral government mandates to prevent users from migrating to less regulated platforms like TikTok. Thompson asked why the go-it-alone approach is deemed appropriate for AI risks, which frontier labs describe as potentially catastrophic, while government intervention is considered necessary for social media harms.
Thompson also examined the business model behind Muse. Zuckerberg described a plan where Meta takes a small cut of transactions facilitated by the agent, effectively acting as an aggregator of consumer demand. Thompson identified this as an aggregation theory similar to Apple's App Store model, where businesses pay fees to reach users. He noted that Meta has historically criticized Apple's 30 percent fees and App Tracking Transparency rules, which he argued stifled Meta's ad business. Thompson questioned how Meta intends to manage friction with downstream businesses who may resent similar fee structures, given Meta's own history as a party aggrieved by Apple's aggregation practices.
Finally, the analysis covered data center investments. Zuckerberg highlighted community benefits from Meta's Hyperion data center in Louisiana, including tax revenue that funded teacher bonuses. Thompson contrasted this narrative with reports of secretive negotiations and property tax exemptions secured by Meta. He suggested that such deals should be conducted with greater public transparency to allow communities to evaluate trade-offs independently. The piece concluded by noting that while Muse targets enterprise applications such as small business commerce, it has yet to demonstrate sufficient consumer utility to overcome widespread skepticism about AI products.