Major Upgrade Launched by Sam Altman's World Project to Combat Deepfakes and Bots

In a significant move, World, a digital identity project backed by Sam Altman, has introduced its most substantial upgrade to the World ID system, positioning it as a comprehensive 'proof of human' infrastructure for individuals, businesses, and AI agents. This overhaul, announced in San Francisco, aims to address growing concerns across the tech industry about bots, deepfakes, and AI agents impersonating humans online by focusing on authentication, payments, and internet services. Altman is also known for his involvement in OpenAI, the company behind ChatGPT and other large language model AI platforms. The World ID system utilizes custom-built 'Orb' devices to establish proof of humanity. To obtain a World ID, users must visit an Orb in person, where the device scans their face and iris to generate a unique cryptographic code. This code is then processed, and only anonymized fragments are sent across a distributed network to confirm the individual has not previously registered, resulting in a credential that proves someone is a unique human online without revealing their identity or personal data. Critics have raised concerns about the use of biometric scanning via the Orb. The update includes a redesigned architecture to enhance privacy, security, and usability, with new features such as account-based identity, multi-key support, and recovery mechanisms. A dedicated World ID app is also being introduced, allowing users to manage credentials and authenticate across platforms. The company is expanding its partnerships with consumer platforms like Tinder, where users can display a 'verified human' badge, and is working with Zoom on a 'Deep Face' feature to verify meeting participants are real humans. Additionally, World is partnering with Docusign to incorporate proof-of-human checks into digital agreements, and with other firms like Okta, Vercel, and Browserbase to establish a trust layer for automated workflows without requiring personal data.