AI-Enhanced OSINT Evidence Governance: Academic Integrity, Platform Disposition, and National Security Risk Assessment in the Case of Shanghai Maritime University’s “First-Class Undergraduate Major” Controversy

This study anchors its analysis on the controversy surrounding Shanghai Maritime University's “First-Class Undergraduate Programs” initiative in China. It aims to validate whether AI-enhanced open-source intelligence (OSINT) evidence governance can reconstruct auditable evidence chains and quantify narrative discrepancies in education-related public incidents, while extrapolating governance performance to multidimensional national security risks. Methodologically, we collected university announcements, mainstream media reports, platform-accessible pages, and judicial leads within a unified time window to construct a “source-time-version” evidence graph. We introduced multilingual NLI and sentence vectors to measure narrative consistency, combined with JSD to monitor monitoring scope drift. We employed temporal propagation networks and DTW to measure the temporal misalignment between “post deletion/restriction” and public opinion peaks. Further, we mapped “evidence governance/information control opacity” to four normalized risk dimensions—academic integrity, rule of law and institutional trust, ideological security, and social stability—using structural causal models, providing 95% confidence intervals. Results indicate: The foundational evidence chain can be robustly reconstructed, yet significant narrative gaps exist regarding “whether ‘7 categories, 40 items’ of fraud occurred” and “whether platform actions were implemented/compliant.” Communication peaks typically precede search peaks by approximately one cycle (moderate DTW misalignment), with public nodes consistently holding the highest intermediary scores. Among the four risk dimensions, “academic integrity” and “ideological security” exhibit relatively higher levels. Conclusions indicate that earlier, auditable evidence disclosure (versioned notifications, third-party verification, platform log retention) simultaneously reduces all four risk dimensions and shortens the public opinion chain, whereas declarative statements alone fail to resolve disputes. AI-enhanced OSINT provides a reusable methodological pathway for evidence governance in educational settings and national security assessments.