Meta Shifts the Blame + Do Data Center Bans Work? + The Final HatGPT

Meta’s $17.1 billion settlement regarding child safety mandates significant product changes, including default two-hour daily time limits and nighttime usage blocks, though the company’s push for competitors like TikTok and YouTube to adopt similar standards is widely viewed as a cynical attempt to avoid unilateral disadvantage. Princeton computer scientist Arvind Narayanan challenges the efficacy of local data center opposition, arguing that industry-wide efficiency gains in hardware and software far outpace the capacity lost by blocking individual facilities. Rather than focusing on infrastructure, Narayanan suggests that better alignment of workplace incentives and individual user control over AI agents are more effective strategies for managing the rapid, often premature, deployment of these technologies. The episode concludes with a final installment of "Hat GPT," covering diverse developments ranging from surgical applications of the Apple Vision Pro to the rise of humanoid robot sprints.