en.Wedoany.com Reported - On April 14, 2026, DTClaw, a professional-grade AI agent under Ant Group Digital Technology, fully passed the "OpenClaw-Class Intelligent Agent Security Protection Product Capability Evaluation" organized by the China Academy of Information and Communications Technology (CAICT), becoming one of the first products in China to pass this evaluation. The DTClaw product is positioned as a "Professional Claw" serving highly specialized users such as financial experts, wealth advisors, and data analysts, clearly distinguishing it from the "Ordinary Claws" currently available on the market that can only handle basic tasks like document organization, meeting minutes, and data collection.
The DTClaw platform has over a hundred built-in professional-grade capability modules and comes pre-installed with numerous mature templates adapted to actual workflows, covering multiple high-value fields such as wealth management, asset allocation, quantitative analysis, software development, and quality assurance. The platform reduces token consumption by over 70% through algorithm optimization while increasing response speed to 2.3 times the industry average. The DTClaw platform integrates financial-grade security mechanisms such as data encryption, access control, and audit logs to ensure the security and privacy protection of sensitive financial data, meeting regulatory compliance requirements.
CAICT organized this evaluation based on the FG-K03-0003-01 "OpenClaw-Class Intelligent Agent Security Protection Product Capability Requirements and Assessment Methods." The evaluation covered three major dimensions: basic environment security, permissions and data security, and application security, comprising a total of 9 testing indicators. Specific test items included core security capabilities such as sandbox isolation, attack surface detection, vulnerability detection, permission control, data protection, plugin security, API security, and prompt injection attack protection. The evaluation focused on the practical security pain points and protection needs in intelligent agent applications, conducting a detailed technical assessment of the product's coverage of protection scenarios, success rate of protection, and impact on business performance.
Prior to DTClaw fully passing the evaluation, Ant Group Digital Technology had already launched the "AntGuard 2.0 - Lobster Guardian" AI security protection system on March 19, 2026, building a multi-layered defense architecture from the content layer, plugin layer, and situational awareness layer. This system successfully intercepted 98.7% of simulated attacks in stress tests, achieved a plugin audit accuracy rate of 99.2%, and reduced risk warning response time to under 3 seconds. Ant Group Digital Technology simultaneously initiated the "Lobster AI Security Guardian Program," offering the first 100 partner enterprises three months of free security protection call services, totaling 200,000 calls.
CAICT has previously announced multiple batches of OpenClaw-class intelligent agent security evaluation results. On April 1, 2026, five products from Tencent, including WorkBuddy, QClaw, and Lightweight Cloud OpenClaw, were among the first to pass the "Cloud-Based OpenClaw Baseline Capability Assessment," which focused on five core issues: functional trustworthiness, cost control, permission reliability, traceability of sources, and manageability of capabilities. On April 2, 2026, Qi An Xin's Lobster Security Companion was among the first to pass the security protection product capability evaluation, becoming the first cybersecurity vendor to be selected. On April 13, 2026, NSFOCUS's AI Security Appliance passed the same evaluation. CAICT also led the drafting of the "Cloud-Based OpenClaw Baseline Capability Requirements" standard, establishing a comprehensive evaluation system covering business quality, rights protection, security protection, open-source compliance, and skill application.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
