This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Google Books has been found to index low-quality, AI-generated books, identified by common chatbot phrases like 'as of my last knowledge update.' This practice threatens the integrity of the Google Books database and the accuracy of the Google Ngram Viewer, a tool used by academics to track language evolution. While Google claims current Ngram data is not affected, the presence of AI-generated content in the index poses a risk to future research and data reliability.
Google Books is indexing low-quality, AI-generated books, degrading its database and potentially distorting Google Ngram Viewer's analysis of language trends. This integration of inaccurate or misleading information undermines trust, disseminates poor-quality content, and wastes resources as researchers must spend time clearing up the misinformation.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Highly personalized AI-generated misinformation creating “filter bubbles” where individuals only see what matches their existing beliefs, undermining shared reality, weakening social cohesion and political processes.
Also classified as
The proliferation of low-quality AI-generated books devalues human authorship and compromises the academic value of published literature.
The indexing of low-quality, AI-generated books pollutes the Google Books database, threatening the integrity of academic research tools like Ngram that track shared linguistic reality.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The incident was primarily caused by humans intentionally publishing AI-generated books to spam the market, and Google's automated systems indexing them.
Added categories peak at 3: Substantial (Epistemic Harm, Direct)
Low-quality, AI-generated books are being indexed by Google Books, threatening the integrity of the database and the Google Ngram Viewer used by academics. While this degrades the quality of linguistic research tools, it has negligible direct national security implications.
The AI system (ChatGPT) acted as a writing assistant to generate text for books, which were then published by humans.
Risk Level 3: The incident involves chatbots (ChatGPT) and AI-generated content (books), which fall under Limited Risk due to transparency and disclosure requirements.
The AI system, ChatGPT, was used to write entire books on various topics, acting as a writing assistant for the publishers.
The report clearly details the discovery of AI-generated books in Google Books using specific search terms and explains the potential impact on the Ngram tool. While the exact number of AI-generated books is not quantified, the mechanism of the risk is well-explained.
None
Technology
Automated Text Generation
Generative AI easily creates low-quality books at scale.
Inadequate Detection Algorithms
Google Books lacks systems to identify AI-generated phrases.
Data Inputs
AI-Generated Book Submissions
Bot-authored books are uploaded directly to Google Books.
Scraped Source Material
AI books use scraped online data containing outdated facts.
Human Factors
Opportunistic Self-Publishing
Authors publish unedited AI content to profit with minimal effort.
Process and Methods
Unvetted Indexing Workflow
Google indexes books automatically without reviewing content quality.
Management
Focus on Index Expansion
Management prioritizes rapid indexing over rigorous data curation.
Delayed Risk Mitigation
Proactive measures to protect Ngram from AI pollution were missed.
Regulatory Environment
Lack of AI Disclosure Mandates
No regulations require publishers to label AI-generated text.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
