the snippet should be written differently.

instead of the ai saying "gruez is japanese" it should say "hacker news alleges[0] gruez is japanese"

there shouldn't be a separate disclaimer: the LLM should tell true statements rather than imply that the claims are true.