The AI 'Death Spiral' Threatening Internet Content Economics

🔑 Enhanced Key Takeaways

•Google's AI Overviews have led to significant drops in click-through rates (CTR) for publishers, with some reporting declines as high as 56% or even 70% in specific industries like fashion, travel, DIY, and cooking.
•The rise of zero-click searches, exacerbated by AI Overviews, is projected to climb towards 70% by mid-2025, significantly impacting publishers who depend on search traffic for ad revenue and conversions.
•Many publishers, including major news outlets like The New York Times, Reuters, and The Associated Press, have filed lawsuits against AI companies such as OpenAI and Microsoft, alleging copyright infringement for using their content to train AI models without authorization or compensation.
•Some AI companies, like Perplexity AI, are attempting to establish new revenue-sharing models with publishers, offering a percentage of ad or subscription revenue when their content is cited in AI-generated answers.
•Regulatory bodies, such as those in the UK, are beginning to intervene, with Google testing tools that would allow website owners to opt out of AI search features, potentially strengthening publishers' bargaining power for licensing agreements.

📊 Competitor Analysis▸ Show

Feature/Company	Google AI Overviews	OpenAI (ChatGPT Search)	Perplexity AI
AI Search Market Share (Jan 2026)	~15.0% (Gemini), AI Overviews appear in ~18% of Google searches	60.7% (AI search market)	5.8% (AI search market)
Weekly Queries	N/A (integrated into Google Search)	250-500 million	~50 million
Impact on Publisher Traffic	Significant CTR drops (e.g., 56% for Mail Online, 61% when AI Overview present)	Contributes to overall zero-click trend, but also sends referrals	Contributes to overall zero-click trend, but also sends referrals
Publisher Licensing Strategy	Has signed deals (e.g., AP, Reddit, Stack Overflow), but also faces lawsuits	Numerous licensing deals (e.g., The Washington Post, News Corp, The Guardian, Financial Times, Axios, Vox Media, Disney)	Launched Publisher Program with revenue sharing
Revenue Sharing with Publishers	Not explicitly a revenue-sharing model; licensing deals vary	Not planning to share advertising revenue with publishers at this point	Offers revenue sharing (e.g., 80/20 split of Comet Plus subscription revenue, or ad revenue share)
Legal Challenges from Publishers	Faces lawsuits (e.g., Penske vs. Google, Chegg vs. Google)	Faces numerous lawsuits (e.g., NYT, Authors Guild, The Intercept, Alden Global Capital)	Faces lawsuits (e.g., CNN, NYT, Chicago Tribune)
Publisher Opt-out Options	Testing an opt-out tool for publishers in the UK	N/A	N/A

🛠️ Technical Deep Dive

Most AI-powered search platforms utilize Retrieval Augmented Generation (RAG) as their core pattern to provide direct answers and mitigate issues like hallucinations and knowledge cutoffs.
The RAG process involves encoding a user's query into embedding vectors, searching an index of precomputed content embeddings (which can include multimodal data), retrieving the most relevant candidates, and then reranking these candidates.
The top-ranked results are subsequently fed into a Large Language Model (LLM) as grounding context for synthesizing the final answer.
AI search systems typically operate at the passage level, focusing on extracting and utilizing short, relevant segments of content rather than entire web pages to directly answer user queries.
Key architectural principles for effective AI search include ensuring content clarity, structured formatting, verifiable jurisdiction, and extractability to facilitate accurate retrieval and generation.
A fundamental principle in these systems is that if insufficient reliable data is available, the system should refrain from generating an answer to maintain accuracy and trust.

🔮 Future ImplicationsAI analysis grounded in cited sources

Content creators will increasingly focus on 'AI-resistant' content and Answer Engine Optimization (AEO).

As AI overviews satisfy direct queries, publishers must create unique, high-effort content (interviews, investigations, proprietary data) that AI cannot easily replicate, and optimize for machine readability and citation to be featured as authoritative sources.

Legal battles over copyright infringement will intensify and shape the future of AI training data acquisition.

Numerous ongoing lawsuits from major publishers against AI companies highlight the contentious dispute over unauthorized content use, and their outcomes will set precedents for how AI models can legally access and utilize copyrighted material.

New, diversified revenue models beyond traditional advertising will become essential for journalistic sustainability.

With declining ad revenue from reduced traffic, publishers will need to explore collective licensing, direct payments for AI model training, and community-centric offerings like events and memberships to ensure economic viability.

⏳ Timeline

2020

Zero-click searches account for 65% of queries, indicating a pre-AI trend of users not clicking through to websites.

2022-11

OpenAI releases ChatGPT, rapidly changing user expectations for obtaining direct answers from AI.

2023-05

Google launches Search Generative Experience (later AI Overviews), integrating AI-generated summaries directly into search results.

2023-12

The New York Times files a landmark lawsuit against OpenAI and Microsoft, alleging copyright infringement.

2024-07

Perplexity AI introduces a revenue-sharing program for publishers, an early attempt at a new compensation model.

2026-06

Google begins testing an AI Search opt-out tool for publishers in the UK, signaling potential regulatory influence and increased publisher control.

The AI 'Death Spiral' Threatening Internet Content Economics

⚡ 30-Second TL;DR

Key Points

🧠 Deep Insight

🔑 Enhanced Key Takeaways

🛠️ Technical Deep Dive

🔮 Future ImplicationsAI analysis grounded in cited sources

⏳ Timeline

📎 Sources (29)

👉Related Updates

Industry Leaders Define New Standards for Podcasting

Data Centers to Consume 20% of US Power by 2035