AI Large Models Team Up with Auditors: DeepSeek and Others Safeguard ARB Project Security

Overview

Recently, the decentralized finance project Arbiter App announced on social media that it has temporarily paused updates. Developers revealed that this quiet period is intended to introduce multiple mainstream artificial intelligence models—including DeepSeek, Claude, GLM, and Qwen—to conduct a large-scale security and audit review of the platform. This move aims to ensure the safety of user assets and lay the groundwork for upcoming new features.

“I haven’t updated in recent weeks because I’m having Astra, Claude, DeepSeek, GLM, and Qwen audit my project for security and safety! I want to make sure your funds are safe (safu)!”

Multi-Model Joint Audit: AI Escorting Financial Security

As generative AI technology matures, artificial intelligence has entered the core fields of code auditing and security testing. For this audit, Arbiter App did not rely on a single vendor but instead chose a multi-model collaborative approach involving DeepSeek. This strategy aims to comprehensively identify potential vulnerabilities through the differentiated logic of various models.

The AI models participating in the audit include:

• DeepSeek: Leveraging its strengths in code understanding and reasoning to deeply scan backend logic.
• Claude: Providing systematic security risk assessments.
• GLM and Qwen: Supplementing reviews from different architectural perspectives to ensure comprehensive, gap-free coverage.
• Astra: Assisting with additional security protection checks.

Technical Strategy and Future Outlook

In the blockchain and decentralized finance (DeFi) sectors, code vulnerabilities often lead to massive capital losses. Traditional auditing methods are time-consuming and costly, while introducing AI for automated preliminary reviews has become a new industry trend. Developers emphasized that this “silence” does not indicate project stagnation but rather aims to forge a more robust security foundation.

Currently, the development team is closely monitoring the audit results from each model. If everything proceeds smoothly, an updated version featuring new functionalities is expected to be released later today. This move not only reflects the developers’ emphasis on user asset safety but also demonstrates the practical application value of large AI models in building trusted digital infrastructure. In the future, similar “human-AI co-audit” models may become the standard workflow for Web3 project launches.