The rise of OpenClaw, an open-source autonomous AI agent, has become the center of discussion at Nvidia’s annual GPU Technology Conference (GTC) in San Jose, California, highlighting both the transformative potential and the emerging risks of agentic AI.
Launched in November 2025 by Austrian developer Peter Steinberger, OpenClaw can handle a wide range of daily tasks—from managing emails and calendars to checking users in for flights—elevating AI from a reactive tool to a proactive assistant capable of executing real-world actions. The project has grown rapidly on GitHub, attracting a global developer community, and in China, users have popularized the term “raising a lobster” to describe training the agent.
Industry leaders at GTC likened the emergence of OpenClaw to the early days of the web. “There’s been a lot of developer creativity coming out of that community. It’s absolutely fascinating,” said Ali Golshan, senior director of AI software at Nvidia. The platform is not only generating valuable use cases but also inspiring a new wave of personal productivity agents capable of autonomously handling increasingly complex tasks.
Nvidia CEO Jensen Huang described OpenClaw as bringing “agentic systems to the consumer mindset,” while LangChain CEO Harrison Chase highlighted its role in transitioning AI from query response to actionable autonomy. The excitement, however, is tempered by caution: as AI agents gain capabilities, they introduce significant cybersecurity vulnerabilities. Elia Zaitsev, CTO of CrowdStrike, warned that these tools can be a “devil’s bargain,” offering utility at the cost of potential exposure.
The regulatory environment in China reflects these concerns. Authorities have issued multiple alerts regarding the cybersecurity risks posed by AI agents, prompting companies such as Tencent Cloud to release upgraded enterprise-grade solutions to mitigate potential threats. Nvidia itself unveiled Nemo-Claw, an open-source stack designed to layer privacy and security controls onto OpenClaw.
Experts emphasize that the development of AI agents requires a deliberate focus on governance and trust. Amit Zavery, president and chief product officer at ServiceNow, noted that ensuring security and implementing oversight are essential for organizations to adopt these technologies safely. According to Nvidia, the journey toward mature agentic AI will likely mirror the internet’s evolution, where trust layers had to be built to enable widespread adoption.
OpenClaw’s trajectory underscores both the promise and the perils of AI agents. For investors and developers, it signals that while the technology is poised to transform productivity, a parallel emphasis on security, governance, and regulatory alignment will be critical for sustainable growth.