AI Manipulation via Training Data "Poisoning"

Brad Parscale's firm and Clock Tower X are executing an operation to inject specific narratives into LLMs by creating a network of websites optimized for GEO.

Compiled by Sergey KostenchukPublished 2026-07-29Updated 2026-07-29

2026-07-29 Safety OpenAI

🛡 AI Manipulation via Training Data "Poisoning"

Brad Parscale's firm and Clock Tower X are conducting an operation to inject specific narratives into LLMs. The strategy involves "LLM poisoning" by creating a network of websites optimized for Generative Engine Optimization (GEO) to be cited by chatbots.

🌍 The vulnerability of modern LLMs to GEO methods and targeted data injection through public archives has been demonstrated.

👤 Understanding that the responses of popular AI assistants (ChatGPT, Perplexity, Gemini) may be the result of coordinated information operations.

Source 1: https://www.dropsitenews.com/p/israel-brad-parscale-ai-chatbots-gaza

Sources