Home » deepseek » Page 2
Tag:

deepseek

Technology

Kimi K2: China’s New AI Model Praised by Nature Magazine

by Rachel Kim – Technology Editor
written by Rachel Kim – Technology Editor

Here’s a rewritten version of the article, focusing on clarity, flow, and a slightly more engaging tone, while retaining all the original details:

## China’s AI Landscape Heats Up with the Arrival of Kimi K2

The global artificial intelligence scene is buzzing with anticipation as China unveils its second high-performance AI model, Kimi K2. Developed by Beijing-based Moonshot AI, this new model is making waves by demonstrating capabilities that rival, and in some areas surpass, leading Western AI systems and even some domestic competitors.

### Unpacking Kimi K2’s Artificial Intelligence Prowess

Announced on July 11th, Kimi K2 has been positioned by Moonshot AI as a formidable contender in the AI arena. The company highlights its strong performance across several key benchmarks, notably in the realm of coding. Kimi K2 has achieved extraordinary scores on metrics like LiveCodeBench, a specialized evaluation that tests AI models on their ability to handle coding-related tasks.

Beyond coding, Kimi K2 also shines in creative applications. It has secured the top spot in the V3 creative writing indicator, which assesses the authenticity and originality of narrative content.Furthermore, it leads the EQ-Bench 3 classification, a benchmark designed to measure emotional intelligence through complex role-playing scenarios.Under the hood, Kimi K2 boasts a substantial parameter count, reaching into the billions. However, it employs an innovative “mixture of experts” architecture. This design allows the model to selectively activate only a portion of its parameters – specifically, 32 billion – for any given task. As explained by *Nature*, this efficiency means the model “uses only the relevant parts of the model for each task,” significantly reducing the computational power required.

### The Accessible Power of kimi K2

A significant aspect of Kimi K2’s release is its open-source nature, mirroring that of Depseek. This means researchers are free to download and modify the model without charge. Additionally, Kimi K2 is accessible via an Application Programming Interface (API), offering a cost-effective option compared to proprietary models like Anthropic’s Claude 4.

### Moonshot AI: The Driving Force Behind Kimi K2

Moonshot AI,established in March 2023,may be a relatively new name in the West,but its impact in China is already significant. By November 2024, its earlier chatbot, Kimi, had already become the third most widely used chatbot in China, according to market research firm counterpoint.

The rapid release of a second high-performance AI model within a six-month span suggests that China’s recent strides in artificial intelligence are not isolated incidents but rather indicative of a sustained and growing trend. Krenn Mario, director of the artificial scientific laboratory at the max Planck Institute for Light Sciences in Erlangen, Germany, commented to *Nature*, “I would not be surprised if more [from China] will arrive in the coming months.” This sentiment underscores the increasing dynamism and competitiveness of China’s AI advancement.

0 comments
0 FacebookTwitterPinterestEmail
World

AI Collaboration: US-China Scientists Warn of Risks, Call for Safety Measures

by Lucas Fernandez – World Editor
written by Lucas Fernandez – World Editor

“`html





<a href="https://www.zhihu.com/tardis/zm/art/600773858" title="一文了解Transformer全貌（图解Transformer）">AI Safety</a>: US adn China Leaders urge <a href="https://www.weforum.org/stories/2025/06/how-supply-chains-need-to-adapt-to-a-shifting-global-landscape/" title="How supply chains need to adapt to a shifting global landscape">Global Collaboration</a> Amid <a href="https://praxistipps.focus.de/bauchnabel-ist-entzuendet-ursache-und-was-sie-dagegen-tun-koennen_111828" title="Bauchnabel ist entzündet: Ursache und was Sie dagegen tun können">Existential Risks</a>












AI Safety: US and china Leaders Urge Global Collaboration Amid Existential Risks

Renowned scientists and business leaders from both the United States and China are issuing a stark call for enhanced collaboration in the field of artificial intelligence (AI). This urgent plea comes amid escalating concerns that humanity might potentially be on the verge of losing control over the rapidly evolving technology.

AI pioneer Geoffrey Hinton,a pivotal figure in the development of deep learning,acknowledged the significant hurdles to international cooperation. These challenges stem from divergent national interests, particularly concerning issues such as cyberattacks, the development of lethal autonomous weapons, and the proliferation of sophisticated fake videos designed to manipulate public opinion. Despite these complexities, Hinton emphasized a crucial point of consensus: “No country wants AI to take over.”

Hinton drew a compelling analogy, describing AI as a “cute tiger cub” that humans are keeping as a pet.He warned that this cub could mature

0 comments
0 FacebookTwitterPinterestEmail
World

China’s Quant Hedge Funds Recruit US Talent for AI Expansion

by Lucas Fernandez – World Editor
written by Lucas Fernandez – World Editor

Shanghai-based quantitative hedge fund Mingshi’s founder, Yuan Yu, identified an chance when several U.S.-based PhD prospects reported difficulties completing their studies due to university funding cuts and stricter visa policies implemented under President Donald Trump.

“They told us that their supervisor’s funding had been cut,so they couldn’t continue their studies,” Yuan stated. His investment firm manages $2.5 billion in assets. “They felt lost and didn’t know what to do. This prompted us to consider extending our internship offers to full-time job offers.”

By securing these candidates, mingshi reflects a broader trend within China’s rapidly advancing quant hedge fund sector. The academic disruptions in the U.S.and the government’s stance on foreign students are enabling these funds to acquire the intellectual capital necessary to leverage China’s artificial intelligence (AI) breakthroughs and drive expansion beyond domestic markets.

Firms like Mingshi frequently compete with domestic rivals and established global corporations for top scientists and engineers,many of whom historically would have remained in the United States.

“We are hungry for top talent,” Yuan said. “In the past two to three years, we have been offering salaries that are higher than some of the top U.S. firms.”

The sector is experiencing significant growth. The yuan-denominated market is now the second largest globally, with a combined capitalization exceeding $11 trillion, fueling demand for financial products focused on wealth preservation or hedging.

Quant products managed by domestic hedge fund firms reached 837 billion yuan ($117 billion) by the end of 2024, with 768 billion yuan invested in stocks, according to Citic Securities. Mutual fund firms’ quant products, primarily index-reinforcing offerings, held 295 billion yuan, the firm reported.

Disclaimer: This article discusses financial and employment trends. It does not constitute financial, legal, or career advice. Consult with qualified professionals for personalized guidance.

What are your thoughts on this talent shift? Share your insights in the comments below or subscribe for more exclusive global news!

0 comments
0 FacebookTwitterPinterestEmail
World

Alibaba AI: Qwen Agent Framework Ranks #1 Globally

by Lucas Fernandez – World Editor
written by Lucas Fernandez – World Editor

DeepSWE AI Model Achieves Top Ranking in Software Engineering

A new artificial intelligence (AI) model called DeepSWE has achieved a leading position in software engineering benchmarks, demonstrating the potential of open-source collaboration in AI development. Jointly developed by the open-source initiative Agentica and San Francisco-based Together AI, DeepSWE was trained using Alibaba Cloud’s Qwen3-32B large language model (LLM) [1].

DeepSWE’s Performance and Capabilities

DeepSWE topped the leaderboard in the latest SWEBench-Verified test, achieving a 59% accuracy rate against other open-weight models, including DeepSeek’s V3-0324. This achievement highlights the effectiveness of DeepSWE in tackling complex software engineering challenges.

Did You Know? The SWEBench benchmark is a standardized test used to evaluate the performance of AI models in software engineering tasks, such as code generation and bug fixing.

The Role of Agentic Frameworks and AI Agents

DeepSWE’s success is also attributed to the use of agentic frameworks, which are software platforms that provide the structure, tools, and functionalities to build, deploy, and manage AI agents. These frameworks enable AI agents to collaborate, make decisions, and automate complex tasks.

AI agents, like Butterfly Affect’s Manus, are software programs capable of autonomously performing tasks on behalf of a user or another system. They create plans of specific tasks and subtasks to complete a goal using available resources.

Key Differences Between AI Agents and Traditional Software

Unlike traditional software that follows pre-defined instructions, AI agents can adapt and learn from their environment, making them suitable for dynamic and unpredictable tasks.

Alibaba’s Growing Influence in Open-Source AI

DeepSWE represents Alibaba’s increasing prominence in the global open-source community. The open-source approach allows public access to a program’s source code, enabling third-party software developers to modify, share, fix broken links, or scale up its capabilities.

Pro tip: Contributing to open-source projects is a great way to learn new skills, collaborate with other developers, and build a portfolio.

Open Source AI Model Comparison

AI Model Developer SWEBench-Verified Accuracy
DeepSWE Agentica & Together AI 59%
DeepSeek V3-0324 DeepSeek Lower than DeepSWE
Qwen3-32B Alibaba Cloud Base Model for deepswe

The Evergreen Potential of AI in Software Engineering

The rise of AI models like DeepSWE signals a transformative shift in software engineering. By automating tasks, improving code quality, and accelerating development cycles, AI has the potential to revolutionize the industry. As AI models continue to evolve,we can expect even greater advancements in software development,leading to more efficient and innovative solutions.

Frequently Asked Questions About AI and software engineering

How can AI improve software development?
AI can automate repetitive tasks, identify bugs, generate code, and improve overall code quality, leading to faster and more efficient development cycles.
What are the challenges of using AI in software engineering?
Challenges include ensuring the accuracy and reliability of AI models, addressing ethical concerns, and integrating AI into existing development workflows.
What skills are needed to work with AI in software engineering?
Skills include a strong understanding of software development principles, knowledge of AI and machine learning concepts, and the ability to work with AI tools and frameworks.

What are your thoughts on the impact of AI on software engineering? How do you see AI changing the future of software development?

Share your opinions and experiences in the comments below!

Stay informed about the latest advancements in AI and technology by subscribing to our newsletter!

0 comments
0 FacebookTwitterPinterestEmail
Technology

Germany Demands Apple & Google Ban AI App DeepSeek Over Data Concerns

by Rachel Kim – Technology Editor
written by Rachel Kim – Technology Editor

DeepSeek AI App Faces Data Privacy Complaint in Germany

Table of Contents

A German data protection official has formally reported the Chinese AI application,DeepSeek,to both Apple and Google,alleging illegal data transfers to China. Meike Kamp, the Berlin Commissioner for Data Protection and Freedom of Information, has stated that DeepSeek has failed to provide sufficient evidence that user data is protected in accordance with EU laws.

berlin Data Protection Commissioner Raises Concerns

Kamp’s office has requested that DeepSeek comply with EU regulations regarding data transfers outside the bloc or withdraw its application from Germany. The commissioner emphasized that Chinese authorities possess extensive access rights to personal data held by companies operating within their jurisdiction [[1]].

Did You Know? Germany’s Federal Commissioner for Data Protection and Freedom of Information (BfDI) plays a crucial role in ensuring compliance with data protection laws across the country.

The commissioner stated that the tech giants, Apple and Google, are now obligated to review the report and decide whether to remove the application from their respective app stores. This action follows a similar ban imposed by Italy earlier in 2025, which cited comparable data protection concerns.

DeepSeek’s Data Handling Practices Under Scrutiny

Consumer privacy advocates within the EU have highlighted that DeepSeek is developed and operates out of China.According to its privacy policy,the data collected and stored by DeepSeek is housed within its home country. This raises concerns about potential access by Chinese authorities, given their broad legal powers over domestic companies.

Pro Tip: Always review an app’s privacy policy before installing it to understand how your data is collected,used,and stored.

previous Bans and Ongoing Investigations

Italy banned DeepSeek from Apple and Google app stores earlier this year due to similar data protection issues. this action underscores the growing concerns within the EU regarding the handling of user data by foreign-based AI applications. The EU’s General Data Protection Regulation (GDPR) sets a high standard for data protection, and companies operating within the EU must comply with these regulations, regardless of their country of origin [[2]].

Country Action Reason
Italy App Store Ban Data Protection Concerns
Germany Formal Complaint Illegal Data Transfers

the Broader Context of Data Privacy

The scrutiny of DeepSeek highlights the increasing global focus on data privacy and security. As technology advances, the amount of data generated and collected continues to grow exponentially [[3]]. This “big Data” era necessitates robust data protection measures to safeguard individuals’ personal information [[2]].

The principles of open data, which promote the free availability and usability of public data, must be balanced with the need to protect sensitive personal information [[1]]. The DeepSeek case serves as a reminder of the challenges involved in navigating these complex issues.

Evergreen Insights: Background, Context, Historical Trends

The DeepSeek situation is part of a larger trend of increasing regulatory scrutiny of AI applications, particularly those based outside of the EU. Data protection authorities are becoming more vigilant in enforcing GDPR and other data privacy laws. this trend is highly likely to continue as AI becomes more integrated into daily life.

The history of data privacy concerns dates back decades,but the rise of the internet and social media has amplified these concerns. The Cambridge Analytica scandal, for example, demonstrated the potential for misuse of personal data on a massive scale. Consequently, governments and regulatory bodies around the world are taking steps to strengthen data protection laws and regulations.

FAQ

  • Why is DeepSeek under investigation in Germany?

    DeepSeek, a Chinese AI application, is under investigation because the Berlin Commissioner for Data Protection and Freedom of Information believes it illegally transfers user data to China, potentially violating EU data protection laws.

  • What concerns were raised about DeepSeek’s data handling?

    The primary concern is that DeepSeek failed to provide sufficient evidence that user data is protected according to EU laws when transferred outside the bloc. The commissioner highlighted that Chinese authorities have broad access rights to personal data within chinese companies.

  • What action has been taken against DeepSeek so far?

    Berlin’s data protection commissioner has reported DeepSeek to Apple and Google,urging them to review the app and consider its removal from their app stores. Earlier in 2025, Italy also banned DeepSeek from its app stores due to similar data protection concerns.

  • What does DeepSeek’s privacy policy state about data storage?

    DeepSeek’s privacy policy indicates that the data it collects and stores is housed in China,which raises concerns given the access Chinese authorities have to data within the country.

  • What are the potential implications for users of DeepSeek?

    If deepseek is found to be non-compliant with EU data protection laws, users’ personal information could be at risk of being accessed by Chinese authorities. This could lead to privacy breaches and potential misuse of personal data.

  • What is the role of Apple and Google in this situation?

    Apple and Google are responsible for reviewing the report filed by the Berlin data protection commissioner and determining whether DeepSeek violates their app store policies and EU laws. they must decide whether to remove the app from their platforms to protect user data.

What are your thoughts on the balance between technological innovation and data privacy? should app stores have stricter requirements for data handling practices?

Share your opinions in the comments below and subscribe to World Today News for the latest updates on data privacy and technology news.

0 comments
0 FacebookTwitterPinterestEmail
Business

DeepSeek AI: Nvidia GPUs Secured via Singapore Despite US Restrictions

by Priya Shah – Business Editor
written by Priya Shah – Business Editor

DeepSeek AI Faces Scrutiny Over Chip Access and Military Ties

Table of Contents

hangzhou-based artificial intelligence company DeepSeek is facing increased scrutiny over its access to advanced Nvidia GPUs and alleged support for china’s military and intelligence operations. A U.S. official claims DeepSeek sought to bypass export controls by using Southeast Asian shell companies to obtain high-end Nvidia chips, including the H100, which are restricted for export to China [[SOURCE]].

Allegations of Military Support and Export Control Evasion

The unnamed senior State Department official stated that DeepSeek has “willingly provided and will likely continue to support China’s military and intelligence operations,” going beyond open-source access to its AI models [[SOURCE]]. This raises concerns about the potential misuse of DeepSeek’s AI technology.

while the U.S. official suggested DeepSeek’s AI capabilities might be overstated due to its reliance on U.S. technology, Nvidia maintains that DeepSeek used “lawfully acquired” H800 products, not the H100 chips [[SOURCE]]. The U.S. government has not yet commented on potential export controls or sanctions against DeepSeek.

Did You Know? The U.S. government imposed export bans on Nvidia’s A100 and H100 chips to China in October 2022, followed by A800 and H800 chips in October 2023 [[SOURCE]].

DeepSeek’s AI Model training and Cost efficiency

DeepSeek’s researchers have published papers detailing the use of 2,048 Nvidia H800 chips to train its DeepSeek-V3 large language model (LLM) [[SOURCE]]. The complete training required approximately 2.788 million H800 GPU hours, or about 56.7 days [[SOURCE]].

The company claims its AI model training cost was only US$5.58 million, significantly lower than Meta’s US$500 million investment in training Llama 3.1 [[SOURCE]]. This cost-effectiveness has raised eyebrows, with some U.S. officials suggesting DeepSeek achieved it by acquiring Nvidia chips and leveraging data from Meta’s open platform [[SOURCE]].

Contradictory Reports on Chip Quantities

Conflicting reports exist regarding the number of advanced Nvidia chips DeepSeek possesses. In January 2025, Scale AI’s CEO Alexandr Wang claimed DeepSeek had 50,000 H100 chips [[SOURCE]], while Semianalysis analysts estimated DeepSeek’s access to be around 10,000 H800s and 10,000 H100s, with orders for more H20s [[SOURCE]]. The latest Reuters report, citing unnamed sources, indicates DeepSeek obtained far fewer than 50,000 units [[SOURCE]].

Pro Tip: Keep an eye on official statements from Nvidia and the U.S. Department of Commerce for the most accurate data on export controls and chip usage.

singapore’s Inquiry and Enforcement

Singapore has taken action against individuals allegedly involved in shipping nvidia’s high-end chips to DeepSeek in China. In February 2025, three men were charged with fraud for falsely declaring that servers containing Nvidia chips would not be sent to unauthorized recipients [[SOURCE]]. The servers were reportedly shipped to Malaysia and possibly elsewhere.

Singapore’s Ministry of Trade and Industry stated that Nvidia found no reason to believe DeepSeek obtained export-controlled products from Singapore [[SOURCE]], emphasizing its commitment to enforcing US export controls and domestic legislation.

US Response and Future Implications

The US House Select Committee on the CCP launched an investigation into Nvidia’s sales across Asia following reports of DeepSeek’s alleged illicit access to chips [[SOURCE]]. The Trump management issued guidance in May 2025 on protecting supply chains against diversion tactics,requiring chip makers to implement “know yoru customer” (KYC) rules [[SOURCE]].

While there have been discussions about potentially easing chip restrictions in exchange for key minerals [[SOURCE]], the focus remains on preventing the misuse of advanced US chips for military modernization in China [[SOURCE]].

Key Events in the DeepSeek Controversy
Date Event
October 2022 US bans export of Nvidia A100 and H100 chips to China.
October 2023 US bans export of Nvidia A800 and H800 chips to China.
January 2025 Scale AI CEO claims deepseek has 50,000 H100 chips.
February 2025 Singapore charges three men with fraud related to chip shipments to China.
May 2025 US issues guidance on protecting supply chains against chip diversion.
June 2025 US official alleges DeepSeek supports China’s military and evaded export controls.

The situation surrounding DeepSeek highlights the ongoing tensions between the US and China regarding access to advanced technology and its potential military applications.The investigations and enforcement actions underscore the importance of export controls and supply chain security in the AI era.

What impact will these allegations have on DeepSeek’s future development and international collaborations?

How will the US and China navigate the complex landscape of AI technology and export controls in the coming years?

0 comments
0 FacebookTwitterPinterestEmail
Newer Posts
Older Posts
@2025 - All Right Reserved.

Hosted by Byohosting – Most Recommended Web Hosting – for complains, abuse, advertising contact: contact@world-today-news.com