In this case it appears it's a "Safety of Meta's relationship with the Commander-in-Chief", it would seem. Nah, wind that back, it's more likely incompetence than malice, as per the old adage. It probably is some kind of filtering oversight, but it's still a bad look and more evidence of the arbitrary and difficult to predict/control nature of AI/LLMs.