{"title":"Can AI-written content rank? What the evidence actually supports","url":"https://outwrite.ai/research/ai-content-quality-evidence","type":"Research review","modified":"2026-10-27T13:00:00Z","methodology":"Focused review of six primary sources: two vendor analyses, two research papers and two Google guidance documents. Sources were inspected on October 27, 2026. We kept populations, dates, detector labels and outcomes separate, rather than pooling incomparable studies. No new ranking or detector experiment was conducted. The chart reproduces one source’s classifications. The approval example and operating framework are our editorial proposals, not tested customer results.","disclosure":"outwrite.ai sells content automation and AI visibility software. This review used AI-assisted discovery, synthesis and independent checks, followed by source verification and editorial revision. It does not test our product or establish ranking, citation or revenue gains.","sources":[{"id":"1","title":"AI-Generated Content Does Not Hurt Your Google Rankings (600,000 Pages Analyzed)","publisher":"Ahrefs / Si Quan Ong","url":"https://ahrefs.com/blog/ai-generated-content-does-not-hurt-your-google-rankings/","published":"July 7, 2025","period":"Collection window not specified","sample":"600,000 available pages from top-20 URLs for 100,000 random keywords","finding":"Detector labels: 81.9% mixed, 13.5% human, 4.6% AI; rank correlation 0.011.","limitation":"Observational ranking sample; detector estimates, not verified authorship or publication success rates.","verified":true,"checked":"2026-10-27"},{"id":"2","title":"Google Doesn't Punish AI Content; It Punishes Bad Content (331k Pages Studied)","publisher":"Ahrefs / Ryan Law","url":"https://ahrefs.com/blog/google-doesnt-punish-ai-content/","published":"July 27, 2026","period":"Ranking sample from June 2026","sample":"1 million top-10 URLs; roughly 300,000 available to crawler; 150,000 with sufficient text for ranking analysis","finding":"5.3% of top-three pages had a 100% AI detector estimate; the top-three share below 50% estimated AI content was 82.2%.","limitation":"Filtered observational sample. The headline's 331k spans different analyses; not the denominator for the ranking percentages. Detector scores are not true authorship.","verified":true,"checked":"2026-10-27"},{"id":"3","title":"Google Search's guidance on using generative AI content on your website","publisher":"Google Search Central","url":"https://developers.google.com/search/docs/fundamentals/using-gen-ai-content","published":"Updated October 1, 2026","period":"Guidance checked October 27, 2026","sample":"Official guidance, not a study","finding":"Calls for manual fact-checking and review of AI-generated content, including metadata, and warns against scaled content without added value.","limitation":"Policy and publishing guidance do not quantify a ranking effect.","verified":true,"checked":"2026-10-27"},{"id":"4","title":"Optimizing your website for generative AI features on Google Search","publisher":"Google Search Central","url":"https://developers.google.com/search/docs/fundamentals/ai-optimization-guide","published":"Living documentation","period":"Checked October 27, 2026","sample":"Official guidance for Google Search AI features","finding":"No ideal page length or required chunking. Search indexing and snippet eligibility matter; serving is not guaranteed.","limitation":"Google-specific guidance, not evidence about every assistant or a promised citation rate.","verified":true,"checked":"2026-10-27"},{"id":"5","title":"GPT detectors are biased against non-native English writers","publisher":"Weixin Liang and colleagues / Patterns; author manuscript on arXiv","url":"https://arxiv.org/html/2304.02819v3","published":"Author manuscript revised July 10, 2023","period":"2023 study","sample":"Seven detectors tested on 91 TOEFL essays and 88 US eighth-grade essays","finding":"Mean false-positive rate on the human-written TOEFL sample was 61.22%.","limitation":"Historical detectors and educational essays, not current Ahrefs validation or website ranking research.","verified":true,"checked":"2026-10-27"},{"id":"6","title":"GEO: Generative Engine Optimization","publisher":"Pranjal Aggarwal and colleagues / KDD 2024","url":"https://arxiv.org/html/2311.09735v3","published":"June 28, 2024 manuscript; KDD 2024","period":"Historical experimental systems described in the paper","sample":"GEO-bench evaluation and tests on two generative engines","finding":"Tested content interventions including citations, quotations and statistics; effects varied by domain.","limitation":"Answer visibility experiment, not Google ranking or conversion experiment; systems and queries can change.","verified":true,"checked":"2026-10-27"}],"chart":{"sourceId":"1","title":"Detector classifications in the 2025 ranking sample","description":"Ahrefs classification of 600,000 available top-20 pages. Estimated labels, not verified authorship. This is not a publication success rate.","unit":"% of sampled pages","rows":[{"label":"Mixed human and AI estimate","value":81.9},{"label":"Human-only estimate","value":13.5},{"label":"AI-only estimate","value":4.6}]}}