Technology
Danish Kapoor
Danish Kapoor

OpenAI claims that Deepseek has trained artificial intelligence models with his own data

In the world of artificial intelligence, the competition is increasingly heating, while the tension between OpenAI and China -based Deepseek gained a new dimension. OpenAI announced that Deepseek has found evidence that it has developed a rival model using its artificial intelligence models illegally.

According to Bloomberg’s news; OpenAI and Microsoft are investigating that Deepseek attracts a large amount of data through OpenAI’s API and trains its own model. Towards the end of 2024, Microsoft’s security team detected large -scale data leaks through OpenAI’s developer accounts. The company thinks that these activities are linked to Deepseek. If the claims are true, this means a data usage that violates OpenAI’s service conditions.

OpenAI said that Deepseek applied the “distillation” method used to produce smaller but efficient models than large artificial intelligence models. This technique aims to create new models at less cost by raising information from the outputs of large language models. According to OpenAI, Deepseek drew information from OpenAI’s GPT models and developed their own systems using this method. However, such use is contrary to the service conditions of OpenAI.

OpenAI argues that this development violates the intellectual property rights of the company, although it does not offer detailed evidence on the issue. Company officials said that OpenAI has developed the world’s most advanced artificial intelligence models and has taken various measures to protect the capabilities of these models.

Deepseek, with the R1 model developed last year, has aroused a great impact in the world of artificial intelligence. The company claims to train this model using only $ 6 million and using 10,000 NVIDIA GPUs. These figures remain quite low compared to billions of dollars of OpenAI and other major technology companies. Deepseek’s development of a powerful model at such low costs created a surprise in the sector.

However, the success of the company is suspected by some experts. In particular, it is a question of how Deepseek can develop such a powerful model without advanced GPUs. In addition, how the company creates data sets and the sources of education are also controversial.

This shows that the artificial intelligence competition between the US and China is increasing. As a US -based technology company, OpenAI is one of the most powerful artificial intelligence models. On the other hand, the Chinese government is working to strengthen its own models by making major investments in the field of artificial intelligence.

OpenAI: We are in close cooperation with the US government

OpenAI claimed that Chinese -based companies are trying to copy the artificial intelligence models of US companies and that they are working closely with the US government. The company said that the most advanced models have taken strict security measures to protect their capabilities and have taken the necessary steps to protect their intellectual property rights.

On the other hand, OpenAI’s claims bring about ethical debates in the world of technology. Because it is known that OpenAI uses a huge amount of data on the Internet without permission while educating its own models. For this reason, some experts consider OpenAI’s criticism as “ironic ..

Danish Kapoor