Mallaby opened with a stark forecast: he puts a 50 percent chance on OpenAI going bankrupt within 18 months. He warned that private firms shouldn’t be the gatekeepers of cutting-edge cyber defense tech. “What happens when China can do this in 6–12 months?” he asked, pointing to the risks of one country controlling vital AI tools. Demis Hassabis spent most of his time on scientific frontiers and how to push them faster. He called Dario Amodei the strongest lab leader in the field. Beyond AlphaFold’s success in protein folding, he said we need six more breakthroughs like it to cut drug-delivery timelines from 10 years to mere months. He also made it clear that the chance of AI causing existential risk isn’t zero—he considers views claiming p(doom) is zero “crazy.” On broader impact, Hassabis painted two sides of an AI-driven world. One side promises a surge in science and innovation; the other poses big economic and ethical puzzles about sharing value. He urged social scientists and economists to join biologists, physicists and philosophers in shaping decisions with decades-long effects. He even noted India’s AI optimism after attending Modi’s summit, then closed with career advice: anyone can tap frontier tools—just build something new with them.