A new benchmark study evaluated over 50 AI models on their ability to detect Kremlin propaganda. The findings indicate that some AI systems were almost twice as likely to repeat propaganda when presented with deliberately misleading questions. This research highlights a significant vulnerability in current AI systems regarding disinformation. AI
IMPACT Highlights AI vulnerability to disinformation, potentially impacting trust and safety in AI applications.
RANK_REASON The cluster describes a new benchmark and study evaluating AI models, fitting the research category.
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →