DeepSeek V4 Pro正式版+Harness上线国家超算互联网
Related
Gu Yuxian, a Tsinghua University Special Scholar, joined DeepSeek, where he previously led the development of large-scale model distillation and a 50x speedup for long text processing.
According to Beating's monitoring, Gu Yuxian, a PhD graduate from the Department of Computer Science at Tsinghua University and recipient of the 2025 Graduate Special Scholarship, has officially joined DeepSeek, and his name has appeared in the author list of the DeepSeek V4 paper. Gu Yuxian's research mainly focuses on efficiency optimization of large models in the pre-training, model compression, and inference stages, and has been cited nearly 5,000 times on Google Scholar. Gu Yuxian's previous representative works include the knowledge distillation method MiniLLM for large models (which has been adopted by platforms such as Google, Alibaba, and NVIDIA), and the hybrid architecture model Jet-Nemotron. Jet-Nemotron achieves a 53.6 times faster throughput than traditional full-attention models when processing 256K ultra-long contexts on an H100 GPU, and surpasses hybrid expert models with larger parameter scales in multiple benchmark tests.
DeepSeek's Secret Self-Developed Inference Chip
According to Beating, Reuters, citing sources familiar with the matter, reports that DeepSeek, a major Chinese model-making company, is developing its own AI inference chip. This research began about a year ago and is primarily for model inference rather than training. In recent months, the company has been privately recruiting chip design engineers. Currently, DeepSeek's model operation and training heavily rely on chips from Nvidia and Huawei. If successful, developing its own inference chip would reduce reliance on external suppliers and provide more cost-effective hardware control. However, the project is still in its early stages and faces manufacturing and memory access restrictions due to US export controls.
After criticizing DeepSeek for disrespectful interviewing practices, Huawei's "genius youth" Li Bojie is accused by a former investor of "taking money and disappearing."
According to Beating's monitoring, Li Bojie, one of Huawei's first batch of "genius teenagers," complained about his DeepSeek interview experience on social media, claiming that he was accused of cheating during a coding interview and therefore terminated the interview on the spot. This complaint quickly sparked a new controversy, with former investor and co-founder of Web3 venture capital firm ABCDE, Du Jun, publicly accusing Li Bojie of "lacking a sense of contract." Du Jun alleges that Li Bojie and his partner Zhuang Siyuan, after receiving investment from ABCDE to found Metagent in 2024, refused to fulfill their contractual obligations to synchronize financial and business progress, and subsequently disappeared. Du Jun stated that he can accept various project failures, but cannot accept the fraudulent behavior of founders taking the money and running away. Li Bojie resigned from Huawei in 2023 and co-founded Logenic AI (later renamed Pine AI), and has not yet responded to these allegations.
Polymarket traders estimate a 23% probability that the US will block public access to major Chinese AI models by the end of 2026.
According to Odaily Odaily, Polymarket traders estimate there is a 23% probability that the U.S. government will block public access to major Chinese AI models by December 31, 2026, through legislation, executive orders, export controls, or other formal actions. Several departments within the U.S. Department of Commerce have already banned employees from using Deepseek on work devices due to data security concerns, and Virginia, Texas, and New York have also imposed bans on Deepseek devices for their state employees. (Bitcoin.com News)
The Cyberspace Administration of China announced the first phase of its campaign to rectify irregularities in AI applications, with over 14,000 non-compliant products dealt with.
According to Mars Finance, the Cyberspace Administration of China recently announced the progress of the first phase of the "Clean Up the Internet: Rectifying the Chaos in AI Applications" special campaign. Since its launch in April 2026, the campaign has focused on issues such as large-scale models failing to register as required, insufficient review and filtering capabilities, AI data poisoning, and inadequate implementation of content labeling. To date, the first phase has dealt with over 14,000 AI products, including illegal websites, applications, and intelligent agents; removed over 6 million pieces of illegal and irregular information; dealt with over 26,000 illegal accounts; and removed over 1,300 illegal AI products and 9 illegal open-source datasets. During the campaign, local cyberspace administrations in many regions have adopted targeted measures such as establishing coordinated regulatory mechanisms and setting up reporting channels. Key platform companies such as Huawei, Alibaba, Zhipu, and DeepSeek have also successively improved their registration review, content interception, and data anomaly detection mechanisms. The Cyberspace Administration of China stated that the next phase of its governance efforts will focus on combating prominent issues such as using AI technology to create and disseminate false information, spread vulgar content, impersonate others, infringe upon the rights of minors, and engage in online troll activities. It will further intensify investigations and urge platforms to improve their prevention and governance capabilities.
Trezor reports data from 14K users exposed through shipping provider
The wallet provider warned that thousands of users could potentially be at risk of phishing attempts, while advising that all devices, private keys and backups were safe.