World has expanded access to AgentKit, a framework that enables AI agents linked to verified World IDs to perform online tasks while helping businesses limit bot activity. According to World, the rollout comes as AI agents take on a growing number of online tasks, increasing demand for systems that can distinguish between software acting for a specific individual and large-scale automated operations. The project, backed by OpenAI CEO Sam Altman, allows users to delegate actions to AI tools while keeping identity verification and user controls in place. The framework is designed to support task delegation while maintaining safeguards tied to verified identity, ensuring that AI activity remains attributable to a real human user rather than anonymous or automated systems.
AgentKit links supported AI agents to a verified World ID, enabling websites and applications to confirm that an agent is acting on behalf of a unique human user. As reported by World, the system is designed to help businesses enforce user-level rules and reduce abuse associated with automated accounts. To access the framework, users need a verified World ID, a World App account, and a compatible AI agent. Supported options currently include Claude Code, Codex, Cursor, Hermes, and OpenClaw through World's ToolRouter interface. Users connect their proof of human through World's ToolRouter interface, generate an API key, and link their AI agent within minutes. Once connected, the agent can interact with services that support AgentKit and perform tasks on behalf of the user.
Virtuals CEO Jansen Teng revealed that the company has expanded beyond gaming-focused AI agents and is now building infrastructure for what it calls an "agent society." According to Teng, Virtuals now focuses on five pillars: creating digital agents, creating physical agents and robots, enabling agent coordination, supporting capital formation and building governance systems for agents. The company's long-term vision centers on a "parallel society" where agents participate in a permissionless economy and collaborate with each other at scale. Teng described the next phase of AI agents as being "less about chat and more about earning, spending and coordinating," with the company building systems where agents can control wallets, trade with one another and perform specialized tasks.
The AI agent framework movement has gained significant momentum in the cryptocurrency trading sector, with Bybit becoming the latest exchange to launch dedicated AI trading accounts. According to FM Intelligence, at least ten retail brokers and platform vendors wired AI agents into live client accounts over the first half of 2026, with most operating on the same open plumbing infrastructure. Bybit's AI Subaccount feature confines all bot activity to a segregated account with no access to main funds, while allowing clients to set leverage caps, maximum allocation limits, and withdrawal restrictions. The exchange provides real-time read-only oversight of AI agents, positioning this as a new standard for risk control in what it calls agentic trading.
The AI agent adoption wave comes amid evolving regulatory frameworks, with Singapore's central bank adding Bybit to its investor alert list alongside Binance and KuCoin this month. While the FCA's first horizon scan flagged AI as a shift it is watching, supervisors including the SEC and ESMA have largely leaned on existing rules rather than creating new frameworks specifically for AI agent trading. The World project, originally conceived by Sam Altman, Max Novendstern, and Alex Blania, aims to provide proof of human, finance and connection for every human in the age of AI. As AI-powered software takes on more responsibilities across digital platforms, World said AgentKit provides a way to connect those agents to verified human identities, allowing organizations to verify who is ultimately behind automated actions carried out online.
