In AI risk contexts, what does Misinformation refer to?

• A. System downtime due to hardware failure
• B. LLMs producing false, credible information often due to hallucinations
• C. Data loss from accidental deletion
• D. Unauthorized access to user data

Answer: (B)

In AI risk contexts, misinformation means the model outputs false information that appears credible, often due to hallucinations. The system generates plausible-sounding statements, numbers, or citations that aren’t grounded in real data, which can mislead users who trust the output. This happens because the model is trained to predict language patterns, not to verify facts, so it may fabricates information when it can’t accurately recall or source it. The risk is about the truthfulness and believability of what the AI presents, not about system uptime, data loss, or unauthorized access. Mitigations include grounding outputs to verifiable sources, adding retrieval-augmented generation, and implementing confidence checks or automatic fact-checking.