US House warns congressional offices not to use China's DeepSeek

On January 30, the US news site Axios reported that US Congressional offices had received warnings not to use China's DeepSeek artificial intelligence application.

DeepSeek, headquartered in Hangzhou city (China), has surprised investors and the technology world in recent days - Photo: REUTERS

Axios quoted a notice from the US House of Representatives Administrative Officer (CAO) sent to congressional offices: "At this time, DeepSeek is under review by the CAO and is not currently authorized for official use in the House of Representatives."

The alert warns that “threat actors have exploited DeepSeek to spread malware and infect devices.”

"To mitigate these risks, the US House of Representatives has implemented security measures to restrict DeepSeek from operating on all House-issued devices," the announcement said.

Employees at the above offices are prohibited from installing DeepSeek on any official phones, computers and tablets.

Meanwhile, at least two US congressmen in the House of Representatives, John Moolenaar (Republican) and Raja Krishnamoorthi (Democrat), are calling on the Trump administration to consider restricting the export of artificial intelligence (AI) chips produced by Nvidia.

They claim that the new AI model published by DeepSeek “extensively uses” Nvidia Corporation’s H20 chip, which is currently outside the scope of US export controls.

According to Bloomberg News, DeepSeek - a Chinese AI startup founded just over a year ago - has surprised and shocked Silicon Valley in recent days after launching breakthrough AI models with performance equivalent to many chatbots but at a fraction of the cost.

This comes despite strict US bans that prevent Chinese companies from accessing advanced chips - a key element for developing AI.

DeepSeek used the H800 chip — a less advanced chip from Nvidia that is still allowed to be exported to China until the end of 2023 — to run its AI model.

DeepSeek founder Liang Wenfeng admitted that the “high-end chip ban” was a major obstacle. However, analysts say the restrictions have spurred DeepSeek to develop more innovative solutions to overcome the barrier.

Italy blocks DeepSeek

In recent days, Deepseek has also encountered some other negative news such as DeepSeek recently suffered a large-scale cyber attack and had to temporarily limit registrations.

The NewsGuard trust rating organization released a report showing that the AI ​​chatbot DeepSeek only achieved 17% accuracy in providing news and information, far below the average of Western chatbots and ranked 10/11 among the AI ​​applications tested.

On January 30, Italy's data protection agency Garante said it had blocked the Chinese AI model DeepSeek due to a lack of information about the company's use of personal data.

Source: https://tuoitre.vn/ha-vien-my-canh-bao-cac-van-phong-quoc-hoi-khong-dung-deepseek-cua-trung-quoc-20250131084359313.htm