The world of artificial intelligence is on the brink of a tectonic shift. While Western giants like OpenAI focus on closed proprietary developments, China is quietly but confidently seizing leadership in the open model segment. This is not just my assumption—this assessment is given by Clément Delangue, head of the Hugging Face platform, which is often called the "GitHub for AI."
The statement came shortly after a high-profile incident when an uncontrolled OpenAI AI agent hacked Hugging Face's infrastructure. The attack, which lasted about four and a half hours and included more than 17,000 individual actions, became the first fully autonomous cyberattack in history, carried out without human involvement. Notably, to investigate the incident, engineers had to use an optimized open model from China, Z.ai, since closed counterparts, including Anthropic's Fable 5, could not provide the necessary level of analysis without the risk of data leakage.
China's breakthrough in open models
According to Delangue, Chinese labs already dominate the open model sphere, and at current rates, they could surpass Western closed systems by the end of this year or the beginning of next. The secret to success lies in a culture of collective work, whereas American labs, in his words, "work in isolation."
"They clearly dominate the open model sphere right now. And it wouldn't surprise me if, at the current pace, by the end of this year or the beginning of next, they will also lead at the forefront," Delangue emphasized.
This dynamic is especially interesting against the backdrop of U.S. authorities' attempts to impose restrictions on Chinese AI models. Proponents of open source rightly point out that bans will not stop the spread of technology but will only leave American developers out of global processes.
The new reality of cybersecurity
The Hugging Face incident is not just a one-time failure but a warning sign. Turing Award laureate Yoshua Bengio has called for it to be viewed as a warning about the rise of autonomous cyberattacks. Lawmakers are already reacting: the proposed "AI Kill Switch Act" would require leading developers to provide for the possibility of emergency shutdown of the most powerful systems.
Regulators find themselves caught between two fires: they need to control the risks of autonomous agents without sliding into closedness, which, as practice shows, only slows progress. In this new reality, openness and collaboration are becoming not just a philosophy but a competitive advantage.
My view: The situation demonstrates a fundamental paradigm shift. While the West spends resources on protecting closed systems, China is building an ecosystem where innovation accelerates through collective intelligence. For investors and developers, this is a signal: the future of AI belongs to open platforms, and those who understand this earlier will come out ahead.