Meta’s oversight committee reviewed the original Threads case and overturned the company’s original decision and first appeal. Regarding a post about outgoing Japanese Prime Minister Kishida Fumio that used a phrase that translates to “go die” in English, the committee determined that the phrase was used figuratively, and not as a literal threat or call for violence.
The incident was triggered by a news article about Kishida and a thread post showing him reacting to his party’s (that) “illegal fundraising.” The caption criticized the prime minister and accused him of tax evasion. In users’ replies, they asked for an explanation from the prime minister, called him a tax evader, and used the phrase “die.” The post also included the expression “ha” and a derogatory term for people who wear glasses. (Watch out, buddy!)
The post received little attention or likes, but someone reported it under Meta’s bullying and harassment rules. Three weeks later, one of Meta’s reviewers decided that the post violated the violence and incitement rules. The user appealed, and another reviewer, like the first, also decided that the post violated the policies. After yet another appeal, the issue was brought to the committee, which accepted the case and overruled the two reviewers who removed the post.
Prime Minister Fumio Kishida (Government of Japan)
“In this case, threats against political leaders use strong language as a non-literal political criticism to draw attention to corruption allegations, and are not uncommon on Japanese social media,” Meta’s oversight committee wrote in an explanation. “Low Likelihood of Harm” The committee considered whether the poster used the word “haa” to determine its figurative meaning.
The committee said the moderator made a “mistake” by removing the post, even though he spoke Japanese and understood the local content. It recommended that Meta clarify its internal guidelines and provide further guidance to reviewers on “how to evaluate language and local content.”
Meta’s oversight committee added that its violence and incitement policy includes a rule banning the phrase “death” to “high-risk people,” but that this is not clear enough. It said its reviewers do not have the authority to evaluate cases involving the phrase “death,” though the company’s policy rationale suggests that context matters in assessing threats. The committee explained to Meta that rhetorical threats using the phrase “are generally permitted except when directed at high-risk individuals,” and reiterated its 2022 recommendation that “provide a standard by which threatening speech directed at a head of state may be permitted to protect rhetorical political speech.”
Additionally, the committee recommended that Meta clarify the difference between its policies for “public figures” and “high-risk individuals.” The committee noted confusion about why threats against public figures are only removed if they are “credible.” In contrast, threats against others are removed “regardless of credibility.”
The watchdog committee, which handed down just 53 rulings last year, has had a busy September: last week it ruled that the phrase “from the river to the sea” should not be banned, and, in a case similar to this one, distinguished between death threats and “expressions of hope” in Venezuela.
