Safeworld raises $12M to test safety of AI-powered robots

Safeworld, a new startup founded by Carnegie Mellon researcher Dr. Ding Zhao, has emerged from stealth with over $12 million in seed funding to develop safety standards for generative AI-powered robots, aiming to prevent accidents as these machines enter real-world environments. The company, backed by Shine Capital and a16z Speedrun, evaluates robotic control systems through simulations populated with realistic human models, testing scenarios like blind corners, tripping, and varied human appearances. This third-party validation is designed to address the unpredictability of generative AI architectures, which lack the predictability of traditional algorithms. Safeworld is partnering with robotics firms like Gritt Robotics, which uses the platform to verify safety for robots working alongside humans in solar farm installations. The founders argue that empirical testing is essential, as formal mathematical proofs cannot guarantee safety, and they aim to establish industry standards before robots become widespread in households and workplaces.