Workers adopt AI faster than employers can support it as humanoid robots break sprint records

Workers across industries are using AI tools at rates that significantly outpace organizational support, creating a governance gap that experts warn could expose sensitive data. Meanwhile, China's humanoid robotics sector continues advancing at a pace that's reshaping expectations for autonomous systems.

The BYOAI problem

A survey by De Nederlandsche Bank found that only 25 percent of employees feel their employers actively support AI use at work, yet adoption is significantly higher. The gap between worker usage and employer support exceeds 15 percentage points in most sectors and approaches 30 percentage points in education and public administration.

This disconnect has spawned what researchers call "Bring Your Own AI" (BYOAI)—employees logging into consumer-grade ChatGPT or Claude with personal accounts on corporate networks. The practice creates blind spots: workers feed proprietary data, client information, and source code into public models without realizing they're forfeiting data privacy.

The largest gaps appear in heavily regulated sectors. Government and education show 30-percentage-point discrepancies because strict data protection laws and bureaucratic processes paralyze institutional adoption, even as individual teachers and civil servants use generative tools to draft lessons and reports. IT, finance, and professional services show smaller gaps, having moved faster to purchase enterprise-grade AI with proper governance.

Humanoid robots outrun humans

China's robotics development hit a symbolic milestone when Tianzhuo, a humanoid robot, ran 100 meters in 9.39 seconds at the World Humanoid Robot Games in Beijing—faster than Usain Bolt's 9.58-second human world record. The robot, developed by Beijing Innovation Centre of Humanoid Robotics, reached a peak speed of 14.5 meters per second.

The achievement follows China's broader push into humanoid robotics as a strategic industry. Unitree Robotics surged 620 percent in its Shanghai market debut, lifting the company's market cap to as high as $66 billion. The Hangzhou-based company shipped more than 5,500 humanoid robots worldwide last year, making it a leading player in commercializing the technology.

Rinat Mirzaitov, founder of Humanoid Analytics, noted that Unitree's market reception reflects its position as "one of the few companies in humanoid robotics with genuine revenue."

U.S. distributor pivots after robot ban

The U.S. response to China's robotics advances has been regulatory. In July, the Federal Communications Commission banned imports of new foreign-made robot models, including Unitree's humanoid robots and robot dogs, citing cybersecurity vulnerabilities. The move was justified on national security grounds but clearly aimed at encouraging domestic production.

RoboStore, the largest U.S. distributor of Unitree products, shut down its distribution business and pivoted to domestic manufacturing. Under the new Robo Inc. banner, the team is developing several products including a humanoid torso on wheels for healthcare, a specialized quadruped for security, and a humanoid platform for educators. They're using 3D printing to build prototypes and expect to produce commercial versions from a new facility.

Split strategies on AI infrastructure

Brazil's approach to AI infrastructure illustrates how countries are navigating U.S.-China tech competition. The government announced 2.3 billion reais ($444 million) in AI investments, splitting projects between U.S. and Chinese firms.

Just over half—1.3 billion reais—will fund a Rio de Janeiro supercomputing project developed with China's Huawei and iFlytek. Separately, about 1 billion reais will go toward a tender for a supercomputer expected to rank among the world's 10 most powerful AI processing machines, likely supplied by Nvidia.

"The strategy is not to depend on a single company, technology or country," the government stated, adding that the investments aim to strengthen national sovereignty over data.

The standardization push

As AI adoption accelerates, the industry is attempting to standardize how agents interact with codebases. Multiple companies are converging on AGENTS.md—a unified markdown file that coding agents can use to understand a codebase. The format is gaining traction with Codex, Amp, Cursor, and others.

The push reflects a broader shift in how development teams work. With coding agents now generating hundreds of commits per day, modularity has become critical. As one developer noted, "The more modular your code, the more agents you can run" in parallel without merge conflicts.

The gap between worker AI adoption and organizational readiness suggests enterprises have significant ground to cover. As one analyst put it: there's a 70 percent market of people who haven't adopted wearables or tracking devices at all—and the same dynamic appears to be playing out with workplace AI, just in reverse.