Peter Steinberger, the Austrian developer who created OpenClaw and now works at OpenAI, told Bloomberg in an interview published this week that the US and China have taken opposite approaches to AI agent adoption — and that America has something to learn from China’s willingness to experiment.
“In the US, I feel that in some companies, if you use OpenClaw, you might get fired,” Steinberger said in the interview conducted at OpenAI’s San Francisco headquarters. “In China, however, it’s the exact opposite in many companies — you might get fired if you don’t use OpenClaw.”
The full interview, conducted by Bloomberg’s Shirin Ghaffary, was translated and republished by PANews on March 28.
Chinese Companies Are Tracking Automation Per Employee
Steinberger described a level of institutional adoption that goes beyond encouragement. Chinese companies showed him internal tracking spreadsheets that list each employee’s name alongside a column for “What was automated today?” — a systematic approach to measuring AI agent productivity gains at the individual level.
“Companies are very proactively encouraging employees to think about how to increase efficiency tenfold,” Steinberger said, according to the PANews translation of the Bloomberg interview.
He acknowledged that neither approach is perfect. The US has seen companies restrict employee access to AI agent tools over security concerns — a pattern also observed at Meta, where Steinberger noted a security researcher was “heavily ridiculed on Twitter” for publicly flagging agent-related issues. China’s regulators, meanwhile, have begun restricting OpenClaw use in state-owned enterprises and government agencies even as the private sector accelerates adoption.
OpenClaw Foundation Weeks Away, Backed by NVIDIA, ByteDance, and Tencent
Steinberger confirmed the OpenClaw foundation is in final stages of formation, with “a few weeks” remaining to finalize the legal and organizational structure. He named NVIDIA as an existing partner, ByteDance as having already joined, Tencent as “in the process of joining,” and Microsoft as being in discussions.
“I’m trying to minimize OpenAI’s involvement because this project needs to remain independent,” Steinberger said. He described the governance model as aiming for “Swiss-style neutrality.”
This confirms earlier reporting from CNBC, which noted that Sam Altman called Steinberger a “genius with a lot of amazing ideas” who would help “drive the next generation of personal agents,” while acknowledging that OpenAI does not own the OpenClaw technology due to its open-source nature.
The Inter-Agent Communication Problem
The most technically specific section of the interview focused on what Steinberger sees as the core unsolved problem for AI agents: secure communication between personal and work agents.
“My agent needs to be able to communicate with your agent,” Steinberger said. He described his own use case: working at OpenAI using Codex for knowledge work daily, but sometimes needing to access data in his personal “claw.” The challenge is building a mechanism that allows cross-agent communication while preventing personal agents from leaking private information to work contexts and preventing corporate data from being exfiltrated to personal devices.
At OpenAI, Steinberger is working on Codex, which he said now has over 2 million weekly users. He argued that the distinction between “programming tool” and general agent will collapse: “All prompts become more powerful with programming capabilities. AI agents are smart enough to know their weaknesses and then compensate for them by writing code.”
Context
The interview adds a founder’s perspective to a story NCT has been tracking for weeks: China’s aggressive OpenClaw adoption, from Tencent’s WeChat integration to grassroots installation events. The specific detail about per-employee automation tracking sheets has not been previously reported elsewhere.