Vietnam recruits influencers and AI experts for propaganda

According to an unpublished draft of the party's propaganda committee seen by Reuters, the Communist Party of Vietnam is proposing an overhaul of its digital information operations. The draft, prepared in April and reviewed by Reuters, calls for building a network of at least 1,000 influencers and 5,000 AI experts by 2030 to disseminate what the document describes as "positive" content. The draft states the strategy seeks to "create 'ideological immunity' for the entire society against harmful, toxic and false information," Reuters reports. It also proposes that at least 80% of Vietnamese-language online content be positive by 2030 and that AI be used to ensure removal within 24 hours of at least 90% of content judged to infringe party guidelines, according to Reuters. The document recommends expanding into formats such as podcasts, short videos and targeted content for younger audiences, and it says AI tools developed by Vietnamese tech companies would help "lead social discussion," as reported by Reuters and regional outlets citing the same draft.

Large-scale information shaping programs that combine human influencers with automated tooling typically rely on two technical components: scalable content moderation pipelines and content generation or amplification systems. Industry-pattern observations: practitioners building such systems often use supervised classifiers and embedding-based retrieval for detection, automated takedown workflows tied to platform APIs for removal, and generative models for template-based messaging and localized multimedia. These approaches can increase throughput but also raise operational challenges common across deployments, including label drift, high false positive rates on borderline content, and adversarial attempts to evade detection.

Observers following state-directed information operations note a recent international pattern where governments pair social-media influence campaigns with AI-driven moderation or generation to shape narratives at scale. Industry-pattern observations: such programs frequently accelerate demand for data engineering, multilingual natural language processing, synthetic-media controls, and vendor contracts for moderation and generative-capability providers. They also tend to create pressure points around dataset provenance, auditing, and the transparency of moderation rules.

Reporting places the draft within a broader shift in Vietnam's power structure, noting Reuters coverage that the country's security apparatus has been expanding influence under party chief and president To Lam. Reuters reports the draft includes numerical targets and short removal windows. For practitioners, the significance lies less in the political judgment of the measures and more in the operational scale being contemplated: coordinating thousands of human agents with automated tooling requires production-grade ML infrastructure, robust monitoring, and workforce pipelines.

For practitioners: monitor public procurement records and job postings for large-scale content-moderation projects, vendor partnerships with Vietnamese tech firms, and legislation that would grant platforms or state actors expanded takedown powers. Industry observers will also watch for published technical requests for proposals, exports or licensing of generative models or moderation tools to Vietnam, and signals from major platforms on how they will respond to state-directed content requests. Finally, track on-the-ground reporting about influencer recruitment and any statements from platforms or companies named in procurement or partnership disclosures.

The factual claims above about recruitment targets, percentage goals, timeframe, and quoted language come from an unpublished draft of the party propaganda committee as reported by Reuters and relayed in regional coverage that cites the Reuters reporting.