Anthropic's warning: AI will start building itself

Anthropic has warned about the next phase of 'recursive AI', where agents could become capable enough to build and train models themselves without human intervention. The idea is that “self-improving” armies of agents could create purely AI-run, zero-person companies that optimise while you’re sleeping. If that’s the story in Silicon Valley, in the UK Katie is at London Tech Week, where everyone from Prime Minister Keir Starmer to AMD’s Lisa Su is focused on tech sovereignty and the question of who owns, controls and shapes AI, not just how fast the technology is advancing. Plus, Ci...