Business and economy·November 4, 2026, 13:00
Chinese AI firm: The race is against the clock, and the code must be shared
AI-generated from the sources listed, with automated source checks.
A DeepSeek engineer defends the company's decision to keep releasing its AI as open source. According to him, sharing speeds up the path to artificial general intelligence, while others argue that open models are hard to secure.

Should advanced artificial intelligence be shared freely, or should access be tightly controlled? The question came up at an AI panel during the World Internet Conference in Wuzhen, China, where DeepSeek engineer Gao Xi defended his company's choice. The source is dated November 3, 2026.
Gao Xi, a research engineer at DeepSeek, said, according to the source, that the company will continue to develop AI as open source. That means others can view, download and build on the work.
Time as the opponent
Gao Xi's main argument is that shared innovations speed up progress toward artificial general intelligence, or AGI. The term refers to AI that can handle a broad range of tasks at the level of a human.
According to Gao Xi, competition in the industry is not about beating other companies but about beating the clock. He also offered an example: If AGI would otherwise arrive in ten years, DeepSeek would like to bring it forward to one year from now. The source renders that sentence in the third person, so the exact wording is uncertain.
What is open weight?
With so-called open weight models, developers can access the code, modify it and run the models locally on their own machines. DeepSeek has, among other things, released an autonomous coding assistant called Super Agent. Alibaba's Qwen team is also betting on openness: According to the source, the team has released more than 460 models with more than 4 billion downloads worldwide.
The concerns
There are skeptics, too. Geoffrey Hinton and Yoshua Bengio have warned in general terms that superintelligent systems could slip out of human control. That is a warning about AI as a whole and is not aimed specifically at open source.
According to the source, critics point out that misuse cannot be prevented once a model's weights have been shared. The weights are the numbers that make up the model's learned knowledge.
OpenAI and Anthropic argue, according to the source, that safety requires controlled access. DeepSeek and Alibaba take the opposite view, arguing that open source benefits everyone.
For ordinary users, the dispute is a sign that the AI tools now emerging can come from two very different schools of thought: closed systems under a company's control or open models that anyone can use and modify.
Source
Get the week's AI news in your inbox
Choose your level, topics and length. One email a week, unsubscribe at any time.
Subscribe to Promptly Newsletter



