World Of AI Heboh: American Programming Equipment Allegedly Using Chinese-made Models

Illustration of the SWE-1.5 model from the Cognition AI, (photo: x @windsurf)

JAKARTA - Two new AI tools for programming from American companies, namely Cognition AI and Cursor, are currently being investigated for allegedly being built on a basic Chinese-made AI model. This controversy restarts debates over openness and lending in the use of open-source models in the artificial intelligence industry.

San Francisco startup Cognition AI, which is now worth around US$10.2 billion (Rp166.5 trillion), has just launched the SWE-1.5 model with programming performance approaching the best level in the world. Although the company said the model was developed from a leading open-source model, they did not explain which model it meant.

Speculations emerged that SWE-1.5 might have been built from GLM-4.6, Zhipu AI's flagship model from China, which was released with an open-source MIT license. Zhipu AI even signaled that Cognition might use GLM-4.6 as its base, although until now there has been no official statement from the Cognition AI.

Another Chinese STARTup Cursor and Trace of San Francisco, with a valuation of 9.9 billion US dollars (Rp161.6 trillion), also faces similar accusations. Its new product, Composer, is claimed to have almost the equivalent of SWE-1.5.

A number of users reported a 'trace' in Mandarin in the results of the AI output, raising suspicions that the basic model also came from China. The Cursor has not yet responded.

Ethics And License Debate

The main problem in this case is not just legal, but ethics. Many Chinese AI models such as GLM-4.6 are released with a MIT license, which allows commercial use and modification without the obligation to include credit. Legally legal, but this practice raises questions about transparency and justice in the world of AI.

According to AI researcher Florian Brand from Trier University, AI's true value is in the process of improvement and re-learning, not just the basic model. In other words, although the basic model is the same, the final result can be very different depending on the advanced training process.

Zhipu AI's Attitude

Responding to this issue, Zhipu AI chose a positive tone. They said open-source model openness actually strengthens the global AI ecosystem. The company claims to have experienced a ten-fold increase in paid users from abroad and plans to expand subscriptions to programming tools to the international market.

For the general public, this controversy shows the importance of trust and transparency in the use of AI. If companies are not open about the origins of the model, users don't know how their data is processed and whether their privacy is maintained.

This case also highlights the ethical dilemma: is it natural that large commercial products are built from open-source works without a clear recognition?

In the end, this polemic shows that AI is now an ecosystem without national boundaries, where technology, data, and ideas exchange across continents. But precisely because it is increasingly global, openness and honesty are the main keys to maintaining user confidence around the world.