Peter Steinberger, creator of the viral open-source AI agent OpenClaw, has joined OpenAI to build user-friendly agents for everyone, including non-techies. OpenClaw started as a fun project called ClawdBot in late 2025, renamed after Anthropic sent a cease-and-desist over its Claude ties and security worries. It exploded in popularity by early 2026 for autonomously handling tasks like browsing, coding, chatting on Telegram or WhatsApp, and managing apps— far beyond just talking like ChatGPT.
Anthropic's tough stance pushed Steinberger away, despite OpenClaw boosting their model's use, while OpenAI snapped up the talent without fully acquiring the project. OpenClaw is now moving to an independent foundation backed by OpenAI, staying open-source amid community skepticism given OpenAI's past.
Today, as of February 2026, OpenClaw thrives with millions of users, GitHub stars, and integrations across platforms, even expanding in China via Baidu and Moonshot AI. LangChain's CEO notes its "unhinged" wild energy drove virality, unlike safe lab releases, signaling enterprises crave secure versions for real work.
Looking ahead, OpenAI eyes dominating agents that do tasks, not just chat, but success hinges on taming security risks for safe enterprise use while keeping the open-source magic alive—lest it fade like past hype. Trajectories point to faster agent adoption if trustworthy, or regulatory stalls if risks persist.
AGENTS RISE, CHATBOTS FALL!