“To mitigate hallucinations, Glean's system verifies each line of an AI-generated answer against the source material provided in the prompt, suppressing or flagging information that cannot be traced back.”