Guardian: 'Poisoned' AI — ChatGPT Shopping Recommendations Steer Users Toward Scam and Fake Websites

The Guardian reports that ChatGPT's shopping recommendation features are being 'poisoned' by adversarial content that steers users toward scam or counterfeit websites instead of legitimate retailers. The pattern follows earlier reporting on AI-optimized content being planted at scale across Reddit and review platforms to influence ChatGPT and Google AI Mode responses; this story documents the downstream consumer-harm phase where those influence operations translate into fraudulent product pages and financial losses for shoppers. The article is framed as an accountability piece establishing that AI shopping assistants are now a live consumer-fraud attack surface, not just a misinformation vector.