AI Daily: Claude Launches Claude Design; DeepSeek Opens First External Funding Round; Ant Group's Lingguang App Flash Application Breaks 30 Million

Welcome to the "AI Daily" section! This is your guide to exploring the world of artificial intelligence every day. Every day, we present you with the latest content in the AI field, focusing on developers to help you understand technical trends and innovative AI product applications.

Fresh AI products Click to learn more:https://app.aibase.com/zh

1、Claude launches Claude Design! One sentence turns into interactive prototype + PPT + marketing materials, and can be packaged directly for Claude Code development

Claude Design is a new experimental product launched by Anthropic, which can generate professional design drafts, interactive prototypes, etc., through text descriptions, images or documents, greatly lowering the barrier for non-designers to create visual content. Its core capabilities include design through conversation, multi-format export, and seamless connection with development, targeting users without design backgrounds and designers.

【AiBase Summary:】

✨ Supports multiple input methods such as text, images, and documents, achieving design through conversation.

📦 Multi-format export and support for one-click packaging and transfer to Claude Code for front-end development.

🎯 Targeting non-designers and designers, accelerating the process from creativity to implementation.

2、Valuation over $10 billion: Domestic large model star DeepSeek starts its first external funding round

As a rising star in domestic large models, DeepSeek has officially started its first external funding round, marking a new stage of resource integration and commercialization in China's AI industry.

【AiBase Summary:】

🧠 DeepSeek has launched its first external financing, planning to raise $300 million, with a post-money valuation exceeding $10 billion.

🚀 Due to increased R&D costs and core talent loss, DeepSeek needs capital support to maintain competitiveness.

📈 The capital threshold in the global AI field is increasing, and domestic funds are concentrating on top companies, showing attention to technological barriers and commercial prospects.

3、Ant Group's Lingguang App has over 30 million flash applications, announces a 100 million yuan special creator incentive plan

Ant Group's AI product Lingguang App announced the launch of the "Lingguang Flash Application Creator Incentive Program", investing 100 million yuan in a special fund to support high-quality content and developers, marking that its "one person, one application" ecosystem has entered a large-scale commercial support phase. Since its release at the end of 2025, Lingguang App has rapidly risen due to its extremely low creation threshold, and the platform has already seen over 30 million flash applications generated by users through natural language, covering multiple dimensions such as tools, games, and social interactions, with the world's largest individual application ecosystem taking shape.

【AiBase Summary:】

📱 Lingguang App launched the "Lingguang Flash Application Creator Incentive Program", investing 100 million yuan in a special fund to support high-quality content and developers.

🌐 Lingguang App now has over 30 million user-generated flash applications, covering multiple dimensions such as tools, games, and social interactions.

💡 Lingguang App launched the public creative community "Lingguang Circle" and conducted in-depth product function iterations, enhancing multi-agent collaboration and full-modal generation capabilities.

4、Moonshot AI partners with Tsinghua University to release PrfaaS architecture, solving the computing power bottleneck of large models

Moonshot AI and Tsinghua University jointly proposed the PrfaaS architecture, optimizing computing power resource allocation by separating pre-filling and decoding, significantly improving the efficiency of large model inference. This architecture achieved cross-regional efficient collaboration, effectively solving hardware limitations in traditional service architectures.

【AiBase Summary:】

🧠 Separating the pre-filling and decoding stages optimizes computing power resource allocation.

🌐 Cross-regional collaboration improves service throughput and response speed.

💡 Dual-time-scale scheduling mechanism ensures transmission efficiency.

5、One image generates a walkable 3D world! NVIDIA opens source Lyra 2.0, thoroughly solving the long video "spatial forgetting" and "temporal drift" problem

NVIDIA Research released the Lyra2.0 framework, solving the spatial forgetting and temporal drift issues in long video generation, supporting the generation of navigable 3D scenes from a single image, with broad application value.

【AiBase Summary:】

🧠 Spatial memory mechanism: maintains 3D geometric information for information routing, avoiding accumulation of geometric errors.

🔄 Self-enhancing training strategy: allows the model to encounter its own degraded output, actively correcting drift.

🌐 Open source and application value: applicable to robot training, game development, and 3D asset generation pipelines.

Details link: https://huggingface.co/papers/2604.13036

6、Ruling! AI secondary creation comics do not necessarily infringe copyright

The Düsseldorf Higher Regional Court ruled that using artificial intelligence to convert others' copyrighted photos into comic style works does not necessarily constitute copyright infringement. This ruling provides new legal basis for the AI creation field, emphasizing the importance of human-identifiable creative decisions in the creation process.

【AiBase Summary:】

📌 The court believes that AI-generated works did not directly copy protected elements of the original work.

📌 The subject of photography itself is not protected by copyright law, only creative elements are protected.

📌 AI-generated works need to have identifiable human creative decisions to obtain copyright protection.

7、Explosion! Anthropic MCP protocol exposed major architectural flaws, dozens of serious vulnerabilities ignite the AI security circle

The article reveals major security vulnerabilities in Anthropic's MCP protocol, which may lead to serious consequences. The vulnerability has been associated with multiple serious CVE numbers and has been confirmed to affect several mainstream open-source projects. Despite repeated communication by the research team, Anthropic refused to fix it, causing controversy in the industry.

【AiBase Summary:】

🚨 The MCP protocol has serious design flaws that could lead to arbitrary code execution (RCE).

💣 It has been associated with 10 serious-level CVE numbers, and the number continues to increase.

🔒 Anthropic refused to fix the vulnerability, calling it "expected design".

8、Anthropic launches desktop pet Claude Buddy, hardware made in Shenzhen

Anthropic's desktop pet Claude Buddy combines hardware with AI, providing developers with a more interesting interaction experience, while also showcasing Shenzhen's strong strength in the field of AI hardware.

【AiBase Summary:】

🧠 Combining the ESP32 development board with Claude Code transforms the code approval process into an interesting interactive experience.

🎨 It has 18 types of ASCII art-style small animal characters, enhancing the fun and emotional value of programming work.

🌍 The M5StickC Plus hardware made in Shenzhen has become an important part of the global AI hardware ecosystem.