Misinformation

Also written: LLM07

An OWASP risk in which the bot states false facts, such as a fake contact address planted in a source. Prompts can't tell a lie from a fact in a source you told the model to trust.