2025-11-05 AI Daily - Anthropic Blocks Chinese Capital, Perplexity vs. Amazon, Google Unveils Space AI Project

The AI field is turbulent today. Anthropic, responding to geopolitical pressure, has begun blocking companies with Chinese capital from using its models, directly impacting platforms under ByteDance. Meanwhile, startup Perplexity publicly accused Amazon of using legal means to suppress its AI assistant. Technically, Google unveiled its “Suncatcher” space AI computing project, demonstrating its ambition to push machine learning to new frontiers.

1. Geopolitical Impact Emerges: TRAE Stops Providing Claude Models

Due to a new policy from its parent company Anthropic, AI service platform TRAE has ceased providing Claude models. This move reflects the increasingly tense geopolitical situation and the rise of technological protectionism.

Behind the Scenes: Geopolitics and Technological Protectionism

The direct trigger for this service interruption was a strict blocking policy announced by Anthropic on September 5, 2025. The policy explicitly states that any enterprise directly or indirectly held by Chinese companies with more than 50% equity, regardless of its registration location, will be prohibited from using Claude series AI services.

Although the TRAE platform is operated by SPRING, a Singaporean subsidiary of ByteDance, it was still included in the blocking list due to its significant Chinese capital background.

Industry analysts believe Anthropic’s decision has two main considerations. First, it responds to geopolitical pressure from the United States, preventing its advanced technology from flowing to countries considered competitors. Second, this move is also a technological protection strategy, aiming to prevent Chinese companies from using “Model Distillation” technology—that is, using Claude as a powerful “teacher model” to train their own AI models with similar performance but lower costs, thereby weakening Anthropic’s market competitiveness.

2. OpenAI Sora App Now Available in More Android Regions

OpenAI announced that the Android version of its video generation application, Sora, is now available in more countries and regions, including:

• Canada
• Japan
• South Korea
• Taiwan
• Thailand
• United States
• Vietnam

Users can download and experience it from the Google Play Store.

3. Claude Code Launches Limited-Time Offer: Up to $1000 Credit

Anthropic brings good news to developers with a limited-time promotion designed for Claude Code, allowing Pro and Max users to explore the powerful features of parallel Claude Code conversations without worrying about exceeding regular usage limits.

• Pro Users: Can receive API usage credit worth $250.
• Max Users: Can receive API usage credit worth $1000.

Event Period: From 9:00 AM PST on November 4, 2025, to 11:59 PM on November 18, 2025.

It should be noted that these credits are only for using Claude Code on web and mobile devices and cannot be combined with other offers.

Official Information: Claude Code Promotion

4. Startup vs. Giant: Perplexity Accuses Amazon of Legal Bullying

AI search engine startup Perplexity recently published a strongly worded blog post, publicly accusing e-commerce giant Amazon of legal threats, intending to prevent Perplexity’s AI assistant from operating on Amazon’s platform.

Perplexity stated in the article that this behavior is “bullying, not innovation,” and believes it is an anti-competitive tactic adopted by Amazon to protect its advertising revenue and control user shopping decisions. Amazon wants users to continue relying on its search system full of ads and sponsored results, rather than allowing users to complete purchases directly through a more efficient AI assistant.

Perplexity emphasized that AI assistants should serve users, not platforms. They stated that they will not be intimidated by such tactics and will continue to fight for user choice and a better online experience.

Official Article: Bullying is Not Innovation

5. Code Understanding Artifact: Cognition Releases Windsurf Codemaps

Cognition, the company that developed the AI software engineer Devin, recently launched a new feature called “Windsurf Codemaps.” This tool aims to solve one of the biggest pain points in modern software development: understanding large and complex codebases.

In the past, engineers had to spend a lot of time reading documentation and tracing code to build mental models. Codemaps automatically generates visual, interactive code maps, allowing developers to quickly grasp the structure, data flow, and dependencies of the code. Whether debugging, refactoring, or taking over a new project, Codemaps can greatly improve efficiency, allowing engineers to truly “understand first, then develop.”

Official Blog: Windsurf Codemaps: Understand Code, Before You Vibe It

6. Anthropic Proposes New MCP Solution to Improve AI Agent Efficiency

Anthropic published a technical article introducing a new method to improve Model Context Protocol (MCP) efficiency through code execution, aiming to make AI agents smarter and more resource-efficient when interacting with a large number of external tools.

Under traditional methods, AI agents need to load all tool definitions into their context window. When the number of tools increases, it consumes a large number of tokens, leading to increased costs and delays. The new solution presents tools as APIs that can be called by code. AI agents no longer need to read all tools at once but can, like developers, explore the file system and only load the tool definitions required for the current task.

According to Anthropic’s data, this method can significantly reduce token usage from 150,000 to 2,000, saving 98.7% of costs and time, significantly improving the scalability and efficiency of AI agents.

Technical Details: Code execution with MCP: Building more efficient AI agents

7. Google’s Sea of Stars: Unveils “Project Suncatcher” Space AI Project

Google unveiled an ambitious research project called “Project Suncatcher,” whose goal is to extend the computational scale of machine learning into outer space.

The vision of this project is to build an interconnected network of solar-powered satellites, each equipped with Google’s own TPU AI chips, utilizing the sun’s endless energy to drive large-scale AI computing. Google believes that AI is a fundamental technology for solving major human challenges, and space will be the next frontier to unleash its full potential.

Currently, Google has begun basic research and plans to collaborate with Planet Labs to launch two prototype satellites in early 2027 to test the performance of its hardware in orbit, laying the foundation for a future era of large-scale space computing.

Project Introduction: Meet Project Suncatcher, a research moonshot to scale machine learning compute in space