As the Metaverse emerges as a pivotal digital paradigm within the Web 3.0 landscape, concerns over user privacy, identity authentication, data integrity, and trust governance are becoming increasingly urgent. This study investigates the design of a multidimensional accountability architecture tailored to the security and trust requirements of decentralized virtual environments. By employing a mixed-methods approach that integrates literature review, case studies of leading platforms (Decentraland, The Sandbox, and Meta Horizon), and a user-centered questionnaire survey, the research identifies critical tensions between decentralization, privacy, and enforceable accountability. The findings reveal stark contrasts between decentralized platforms—characterized by user autonomy, on-chain governance, and anonymity—and centralized platforms that prioritize strong identity systems, real-time content moderation, and institutional oversight. Through a structured analysis using a five-layer security model and SWOT framework, the study proposes a hybrid architecture that balances verifiability, privacy preservation, and governance transparency. It also recommends integrating technologies such as zero-knowledge proofs, decentralized identity (DID), and AI-driven auditing tools to enhance accountability without undermining user sovereignty. Ultimately, this research provides both theoretical foundations and practical design pathways for building secure, transparent, and trust-enhancing metaverse platforms aligned with Web 3.0 principles.