AI Daily: Alipay Launches AI Glasses for Scanning Fortune; Kunlun Wanzhi Releases Tian Gong Skywork Desktop Version; OpenClaw Exposed with Severe Vulnerability

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 technology trends and learn about innovative AI product applications.

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

1. Alipay Launches AI Glasses for "Collecting Blessings" for the First Time

The article details Alipay's AI glasses "Collecting Blessings" feature launched during the 2026 Spring Festival. This feature improves user experience through hands-free operation and smart interaction, expanding the scope and rewards of collecting blessings.

AiBase Summary:

👁️‍🗨️ Hands-free operation: Users just need to wear compatible AI glasses, and the glasses can automatically recognize any "blessing" character they look at.

🤖 Smart interaction: Combined with simple voice commands, the glasses will automatically complete the scanning logic and send the blessing card to the user's Alipay account.

🛍️ Enhanced gameplay: The recognition range is broader, including not only traditional "blessing" characters, but also creative "horse" characters and various popular game characters.

2. Kunlun Wanzhi Launches "Tiangong Skywork Desktop Version": Building the "Strongest AI Brain" for Personal Computers

Kunlun Wanzhi Technology Co., Ltd. has released a new desktop AI application called "Tiangong Skywork Desktop Version." This application redefines the intelligent definition of desktop office work through its powerful local processing capabilities. Its core highlights include local execution without relying on the cloud, top model selection, and full format coverage, providing a new optimal solution for handling large files and high privacy office scenarios.

AiBase Summary:

🧠 Local Execution Without Relying on the Cloud: All tasks are performed in a local virtual machine, ensuring that users' sensitive files stay on their devices.

⚙️ Top Model Selection: Users can freely switch between globally top models such as Claude Opus4.5, Claude Sonnet4.5, or Gemini3Pro based on task requirements.

🎨 Full Format Coverage: Supports Windows systems and can deeply process complex file formats such as images, videos, and tables.

3. Anthropic Plans to Repurchase $35 Billion in Equity, Allowing Employees to Cash Out Part of Their Shares

Anthropic announced a major plan to conduct an employee equity buyout with a valuation of at least $35 billion, while also seeking to raise up to $2 billion in funds. This move not only allows employees to cash out part of their shares but also indicates the company's confidence in its future development.

AiBase Summary:

💰 Anthropic plans to conduct an employee equity buyout with a valuation of at least $35 billion, allowing employees to cash out part of their shares.

💼 The company plans to raise up to $2 billion in funding in the next round of financing, further promoting business development.

📈 This move is considered to help increase the company's overall valuation and lay a solid foundation for long-term development.

4. Ant's Afu Upgrades "Elder Mode": Large Font, Can Call Directly for Consultation

The Ant Afu App upgraded the "Elder Mode," optimized the interface design, enhanced the voice interaction function, and supported custom digital human images, further improving the user experience for elderly users.

AiBase Summary:

📱 Afu App launches "Elder Mode" with large font and icons for easier use by elderly users.

🗣️ Supports voice conversation and dialect recognition to enhance interaction with Afu.

🖼️ Users can customize the "Afu" digital human image, with more styles to be added in the future.

5. OpenAI's Flagship Model Gets Major Update: GPT-5.2 Series Increases Reasoning Speed by 40%, Prices Remain Unchanged

OpenAI announced that its latest flagship AI model, GPT-5.2 and GPT-5.2-Codex, has increased reasoning speed by 40%. This improvement comes from optimizing the reasoning stack, while keeping the same model architecture and weights. This update reduces latency and increases response speed, applicable to all API usage ranges, and developers do not need to make any adjustments. Performance improvements come with unchanged prices, offering higher cost-effectiveness for users.

AiBase Summary:

🧠 GPT-5.2 and GPT-5.2-Codex have improved reasoning speed by 40%, optimizing the reasoning stack to reduce latency.

🚀 Maintains the same model architecture and weights, with no changes in underlying capabilities or training data, so developers don't need to adjust anything.

💰 Performance improvements come with unchanged prices, offering higher cost-effectiveness for users.

6. Apple Joins OpenAI to Launch Xcode Intelligent Agent, Supporting GPT-5 Level Models

Apple launched Xcode26.3, marking a major leap from 'AI Assistance' to 'Intelligent Agent Coding' in its integrated development environment (IDE). The new version introduces native intelligent agent tools, allowing developers to directly call Anthropic's Claude Agent and OpenAI's Codex for complex automation programming tasks. Intelligent agents can not only write code but also fully access Xcode's internal functions, achieving engineering-level cognition and full-process automation. In addition, Xcode26.3 supports the Model Context Protocol (MCP), making it an open 'intelligent agent niche.'

AiBase Summary:

🧠 Introduces native intelligent agent tools, supporting calls to Anthropic's Claude Agent and OpenAI's Codex for complex automation programming tasks.

⚙️ Achieves a transition from AI assistance to intelligent agent coding, where intelligent agents can independently explore project structure, metadata, and API relationships.

🛠️ Supports natural language-driven "one-click" development, where intelligent agents can automatically look up documentation, break down tasks, write code, and run tests.

7. Be Alert! Popular AI Intelligent Agent OpenClaw Exposed with Serious Vulnerabilities, macOS Users Face Virus Risks

The article reveals the security risks of the AI intelligent agent OpenClaw, pointing out that attackers are using its vulnerabilities to launch malicious software attacks on macOS users, bypassing system security protections and risking sensitive data. Experts warn users to be cautious of non-official integration scripts to prevent loss of control over permissions.

AiBase Summary:

🚨 Security Alert: The popular AI intelligent agent OpenClaw has serious vulnerabilities, and attackers are using them to spread spyware to macOS users.

🧬 Attack Method: Hackers use fake "skills" files to trick users into executing malicious Shell commands, forcing the shutdown of macOS's security protection mechanisms.

⚠️ Data Threat: The virus focuses on silently stealing passwords, session Cookies, and developer API tokens, which may lead to full penetration of personal and enterprise cloud assets.

8. WeChat Issues Spring Festival Marketing Rectification Notice: Tencent Yuanbao Was Temporarily Banned for "Inducing Sharing"

WeChat officially issued a notice regarding the marketing activities of Tencent's AI app "Yuanbao," penalizing it for disrupting the platform's ecological order through inducements to share.

AiBase Summary:

🍎 Induces users to share to WeChat groups through "task completion" and "red envelope collection."

🚫 WeChat took measures against "Yuanbao" App's violation links according to regulations.

📢 WeChat emphasized that it will continue to monitor marketing activities and strictly enforce platform rules.