France and other countries investigate their data practices


Italy was the first to block access to AI DEPEEK R1 through Apple App Store mobile stores and Google Google. And it seems that he will not stay in the only one, since Taiwan has just banned China for security reasons that have to do with their origin.

An AI that has been boastful by several of the CEOs that are pushing half the world towards generative artificial intelligence, such as Microsoft Nadella or Sam Altman of OpenAi that have shared the same positive adjectives, but which begins to be in the goal by him possible treatment I would be giving to the data of users through their mobile apps and desktop version.

South Korea, France, Italy, Ireland and other countries They are investigating the practices of the data of a major language model that has disrupted the rules of the game of technological giants that have invested large amounts of capital in their own models; especially for its great efficiency, its “deep thought” model, be Open Source (Your code is in view of all) and for being able to move to GPT-4O in some Benchmarks.

As has happened with the social network of Tiktok videos – which due to its origin and the supposed treatment it has given to the data, even with servers in the United States – the AI ​​DEPEEEK R1 begins to receive the attention of the government agencies of half world.

United States, according to Big Breaking Wireright now he is investigating the Chinese startup about the possibility that he has acquired advanced nvidia semiconductors Through third parties in Singapore, thus avoiding US export controls. The key is in the prohibition of the export of certain Nvidia high performance components: the H100 and A100 GPUS. They are essential for the development of artificial intelligence.

Liang Wenfeng, CEO and co -founder of Deepseek.

Liang Wenfeng

Omicrono

Ongoing research It is formed by key White House and FBI agents, which would be examining whether the Startup Deepseek would have used third -party channels to avoid restrictions. The development of an AI that has been able to overcome GPT-4O to a fraction of the cost has raised doubts about whether it has really hidden certain details such as the use of those Nvidia components; Even Openai, surprisingly, states that Depseek has been trained with chatgpt.

Liang Wenfengfounder of Deepseek, in an interview given to The China Academyindicates the reason for its commitment to efficiency in AI: “Our goal is AGI (general artificial intelligence), which forces us to explore new models structures to achieve higher capacities with limited resources. Apart from architecture, we have studied the ‘Data healing’ and the human reasoning capacity in all our models “.

Nvidia, on the other hand, has declared that from Singapore They have no clue On the possibility of distribution of its chips to third parties in China, although the company has avoided commenting directly on ongoing research.

Leave a Reply

Your email address will not be published. Required fields are marked *