Research Shows LLMs Vulnerable to "Grooming" Attacks That Exploit Poor Reasoning to Spread Falsehoods

Summary

Research reveals that generative AI chatbots lack the reasoning capabilities needed to counter "LLM grooming" — the mass-production and duplication of false narratives online designed to manipulate LLM outputs. Even the latest "reasoning" models remain vulnerable to this form of manipulation, where bad actors exploit poor reasoning in AI systems to produce and spread propaganda and falsehoods at scale.