CNA's editor: AI spotted election manipulation we never asked it to find

Walter Fernandez doesn’t mince words. The Editor-in-Chief of CNA, the Singapore-based network reaching 150 million homes, calls his newsroom “all in” on AI—a stance he’s held since 2019, long before ChatGPT made the technology a household name. But his aggressive vision comes with guardrails. The team spent a full year drafting AI guidelines that explicitly ban cloned voices and AI-generated footage in news coverage. No vanity projects allowed. Everything must solve a real pain point.

That pragmatic approach is why CNA now has over 20 custom GPTs in production. The turning point? Parliament. Covering legislative sessions was grueling, so the team built “Parliament AI” to recognize over 90 MPs, transcribe speeches, and generate searchable summaries. When journalists saw the tool actually work, resistance evaporated. A general-purpose “Newsroom Buddy” GPT—which helps brainstorm ideas and checks copy against the CNA style guide—has become one of the most popular tools. Over 500 enterprise licenses have been deployed across CNA, with another 2,000 at the group level.

During Singapore’s recent General Election, Fernandez saw something that genuinely surprised him. Reporters used ChatGPT as a “second brain,” feeding it verified information for context. But the real magic came from OpenAI’s reasoning models, which analyzed social media campaigns for manipulation. The AI flagged a connection between two suspicious accounts that had changed their profile names mid-campaign. “We hadn’t asked it to look for that, nor had we considered it,” Fernandez said. The system surfaced an anomaly in real time and helped uncover hidden links.

Fernandez’s message to other newsrooms is blunt: the “wait and see” phase is over. But he’s not selling an AI utopia. In a world drowning in what he calls “AI slop”—infinite content generated in minutes—he believes differentiation will come down to quality and relevance, not format or language. The North Star stays fixed on public service journalism. AI is just the tool to get there faster.