Russia benefitting from Data Voids to influence LLMs (www.disinformationindex.org)

A recent report from the Global Digital Institute (GDI) reveals how Large Language Models (LLMs) like Claude Sonnet and Mistral Large are being harnessed by Russian state-affiliated entities to propagate false narratives, significantly influencing public perception. With zero-click searches now making up 60% of online queries, those controlling AI-generated summaries effectively shape narratives that lack necessary context and critical rebuttals. Historical precedents of fabricated narratives highlight the dangers of misinformation, which can now be amplified by AI technologies, potentially leading to widespread belief in these distorted stories. This situation raises urgent concerns for the AI/ML community, as it underscores the dual-edged nature of powerful AI tools. While LLMs offer innovative capabilities for information dissemination, they also present risks when misused for propaganda. The report suggests that AI developers and users can implement measures to mitigate these dangers, emphasizing the need for responsible AI deployment and enhanced context in AI summaries. As AI continues to play a pivotal role in information access, addressing these vulnerabilities is crucial to preserving truth and combating misinformation.