An AI assistant deployed for internal employee use begins producing outputs that reference confidential merger discussions after being connected to the company's document management system. What type of AI risk does this represent?

B is correct because the AI assistant is retrieving and surfacing confidential merger documents from the document management system — indicating that retrieval access controls are insufficient and confidential information is crossing its intended access boundary. A requires evidence the information is fabricated rather than real. C requires evidence of external attack. D requires evidence of malicious prompt injection.