BBC journalist poisoned ChatGPT and Gemini with a single blog post in 20 minutes; Google updated spam policies but manipulation is ongoing at scale.
Key Takeaways
A single well-crafted blog post published almost anywhere can redirect AI Overviews and ChatGPT answers within 24 hours.
Manipulation targets high-stakes queries: health supplement safety, retirement financial advice, and local business recommendations.
Google’s policy update officially bans AI manipulation attempts; caught sites face removal or downranking from search entirely.
Mitigations observed: self-promoters quietly removed from AI answer lists, more uncertainty labels, explicit spam-filtering disclosures in ChatGPT and Claude responses.
As Google cracks down on blog posts, manipulators shift to paying YouTube influencers – and Google’s AI now cites YouTube videos.
Hacker News Comment Review
Commenters frame this as a new SEO cycle (already branded GEO – Generative Engine Optimization) with no structural end, just evolving exploits and patches.
Debate on severity: some note the demo used an obscure novelty query, while others point to real-world astroturfing on Reddit and Wikipedia as evidence the threat is already systemic.
Google’s decades of spam-fighting infrastructure (PageRank, ML signals) exists but appears underutilized here, which commenters read as a leadership failure rather than a technical one.
Notable Comments
@simonw: Confirms personal index poisoning still live 18 months later – Google AI still names a fake whale identity he planted in 2024.
