Wednesday, March 18
AI Collides With Reality
The Pentagon (opens in a new tab)'s relationship with AI companies is fracturing as national security concerns override commercial partnerships. The Justice Department (opens in a new tab) now openly states that Anthropic cannot be trusted with warfighting systems, forcing the Pentagon to develop alternatives to the company's Claude models for military applications. This distrust extends beyond technical capabilities to fundamental questions about AI companies' reliability in critical government operations, marking a significant shift from the previous embrace of Silicon Valley AI partnerships.
Corporate executives are discovering that AI cannot replace sound legal judgment, often with expensive consequences. The CEO of Krafton (opens in a new tab) learned this lesson the hard way after asking ChatGPT how to void a $250 million contract, ignoring his lawyers' advice, and subsequently losing terribly in court. Meanwhile, researchers have documented 126 cases (opens in a new tab) of AI-induced psychological harm since January, creating a growing body of evidence about the technology's unintended consequences on human wellbeing.
Big Tech platforms are making strategic retreats from ambitious projects as market realities set in. Meta (opens in a new tab) is shutting down Horizon Worlds on Meta Quest, effectively ending its most visible metaverse experiment after years of investment and hype. Google expands its Personal Intelligence (opens in a new tab) feature to all US users, betting that personalized AI services will prove more commercially viable than virtual worlds. These moves reflect a broader industry pivot toward AI applications with clearer user demand and revenue models.