TESTING COMPETITOR SABOTAGE IN AI RESPONSES

Summary from the TLDR Marketing newsletter

Negative GEO is possible because some AI models will surface false negative claims if that content appears on discoverable third-party websites. In a controlled test using a fake persona and 10 established sites, only 2 of 11 models cited the planted content after several weeks. Perplexity included the negative allegations and treated site citations as enough support, even without mainstream sources. ChatGPT referenced the same sites but flagged credibility issues and noted the lack of corroboration from trusted outlets. Results show that model trust checks and source authority strongly affect whether damaging claims appear in AI answers. This tactic does not scale easily across most models.