A subtle but significant shift occurred in how large language models (LLMs) cited information sources. Domains that had long dominated AI responses, most notably Reddit and Wikipedia, saw a sharp decline in citation frequency within ChatGPT, while remaining comparatively stable across other AI platforms such as Google AI Mode and Perplexity.
For SEO, AI visibility, and content strategists, this moment deserves closer examination. It was not simply a platform quirk or a temporary anomaly. Instead, it highlighted the growing volatility of AI-driven discovery and the evolving rules that determine which sources LLMs trust, surface, and reference.
This article explores what changed, what the data reveals across platforms, and how brands should respond as AI citations become an increasingly important visibility signal. All with the help of Semrush’s data.
What Changed in AI Citation Behaviour?
Throughout early August 2025, Reddit and Wikipedia were cited in a substantial proportion of ChatGPT responses, appearing in over half of prompts in some datasets. By mid September, that presence had collapsed dramatically. Reddit citations fell from nearly 60% of responses to close to 10%, while Wikipedia dropped from around 55% to below 20%.
Initially, many practitioners pointed to a coinciding technical change in Google Search, specifically the removal of the num=100 parameter, as a possible cause. However, further analysis suggests that this correlation does not fully explain the scale or isolation of the decline.
Importantly, the drop was largely confined to ChatGPT. On Google AI Mode and Perplexity, citation patterns for both Reddit and Wikipedia remained comparatively stable, indicating that this was not a systemic shift across all LLMs.
What became clear is that AI visibility is not governed by a single universal logic. Each platform applies its own weighting, trust signals, and retrieval strategies, leading to materially different outcomes for the same sources.
A Platform-Specific Reality
The divergence between platforms underscores a key lesson for marketers: AI discovery is fragmented.
On Google AI Mode, user generated content (UGC) platforms such as YouTube, Reddit, and Facebook gained share during the same period, while domains like Medium, Quora, and LinkedIn experienced declines. Perplexity showed far less volatility overall, with only modest shifts in citation frequency, and a gradual downward trend for Reddit.
ChatGPT, by contrast, exhibited the most abrupt change. While Reddit and Wikipedia declined sharply, other publishers benefited. PR Newswire, Forbes, and Medium emerged as notable winners in ChatGPT citations following the September change, suggesting a rebalancing rather than a wholesale reduction in external references.
This pattern reinforces the idea that citation visibility is not static. It is shaped by platform decisions that may prioritise authority, freshness, format, or perceived reliability in different ways over time.
Winners, Losers, and the Role of Authority
One of the most revealing outcomes of the study was that traditional “authority” alone did not guarantee stability. Reddit and Wikipedia remain among the most cited domains overall, even after the decline, yet their volatility exposed how quickly AI trust signals can shift.
Meanwhile, LinkedIn demonstrated steady growth across platforms. On Google AI Mode in particular, LinkedIn appeared in close to 15% of responses, suggesting that professional, authored content may be gaining favour in certain AI contexts. Forbes followed a similar upward trend, benefiting from its combination of editorial structure, topical breadth, and recognisable brand authority.
These movements suggest that AI platforms may increasingly differentiate between types of authority: knowledge driven by community, institutional reference material, and professionally authored content may each be weighted differently depending on the query and platform.
Implications for SEO and AI Visibility
For search and content teams, this moment serves as a reminder that AI visibility is inherently volatile and cannot be approached with static assumptions.
Visibility Beyond Rankings
As AI responses become a primary interface for information discovery, being cited matters as much as being ranked. However, citation presence does not map cleanly to traditional SERP positions. Domains with millions of rankings can still lose AI visibility overnight if platform signals change.
This reinforces the need to monitor specific to AI performance separately from conventional SEO metrics.
Platform-Specific Optimisation
A “one size fits all” AI strategy is unlikely to succeed. Brands need to understand:
- Which domains dominate citations in their specific topic areas
- How those patterns differ across ChatGPT, Google AI Mode, and Perplexity
- Where competitors are gaining or losing ground
Without that platform-level insight, optimisation efforts risk being misaligned with how discovery actually occurs.
What Marketers Should Do Now
Rather than reacting defensively to volatility, the opportunity lies in analysis and adaptation.
Marketers should:
- Audit which domains and content types are most frequently cited in their niche
- Track AI citation trends over time, not just absolute presence
- Identify gaps where competitor content is shaping AI responses
- Invest in content that aligns with the formats and trust signals favoured by each platform
The goal is not to “chase” every fluctuation, but to build resilience by understanding how and where AI systems source their information.
Conclusion
The sudden decline in Reddit and Wikipedia citations on ChatGPT was not merely a curiosity; it was a signal. It revealed how fragile AI visibility can be, and how dependent brands are on opaque, platform-specific systems that continue to evolve rapidly.
As AI becomes a more prominent gateway to information, marketers must move beyond assumptions rooted in traditional search behaviour. Visibility is no longer just about rankings or traffic, but about influence within AI generated answers.
Those who invest early in understanding citation dynamics, platform differences, and content trust signals will be better positioned to adapt as AI-driven discovery continues to mature.
At GA Agency, we help brands navigate the changing intersection of SEO, AI visibility, and digital authority. From AI citation analysis to content strategy and data led optimisation, we work with forward-thinking teams to build durable visibility in an AI first world.
Let’s talk about how your brand can stay visible as discovery continues to evolve. Contact us today to get started.
Article author
Chiara Riva
Business Development Assistant
