Slide 27 of 27
CompleteSlide 27
Slide 27 · Complete
LLM09:2025 Complete
You’ve covered all 5 parts, read the real incidents, and tested yourself.
What misinformation is — both the model’s half (hallucination) and the user’s half (overreliance)
Why OWASP merged Overreliance into Misinformation in the 2025 edition
How hallucination happens mechanically — pattern prediction, no ground-truth lookup, confidence as a style
4 misinformation types — each anchored to a real, confirmed incident
Mata v. Avianca (2023) — fabricated case citations, court sanctions, the overreliance loop
Slopsquatting (2024) — how hallucinated package names become a supply-chain attack vector
Moffatt v. Air Canada (2024) — the ruling establishing legal liability for LLM chatbot misinformation
Pearce et al. (2022) — 40% of Copilot’s security-sensitive code suggestions were vulnerable
All 3 OWASP attack scenarios, each grounded in a real-world example
6 mitigation categories — what OWASP says, how real incidents showed the gap, how to do it right, how to validate
Review from beginningNext: LLM10 →
📄 View all sources for this lesson