This seems DOA because how could you possibly prove that an AI “knowingly” wrote false statements. We have to figure out if our current systems have consciousness before we can declare the bot not only knew the correct version of what it could have said but also that it knowingly deviated from that and published falsehoods.