Kenya’s National Cohesion and Integration Commission (NCIC) has warned Facebook to either stop hate speech on its platform within seven days or face ban, as the country goes to election next month.
Meta, the parent company of Facebook, said last week it has been preparing for the country’s 2022 election over the past year with the help of a dedicated team that’s working closely with election authorities and trusted partners in the country.
However, local advocacy group Global Witness and legal non-profit Foxglove have discovered that Facebook “appallingly failed to detect hate speech ads in the two official languages of the country: Swahili and English”.
Reacting to their report, the country’s ethnic cohesion watchdog NCIC said that Facebook is in violation of the laws of the country.
“They have allowed themselves to be a vector of hate speech and incitement, misinformation and disinformation,” NCIC commissioner Danvas Makori said in a statement.
Global Witness and Foxglove said this follows a similar pattern they uncovered in Myanmar and Ethiopia, “but for the first time also raises serious questions about Facebook’s content moderation capabilities in English”.
“Facebook itself has praised its asuper-efficient AI models to detect hate speech’ but our findings are a stark reminder of the risk of hate and incitement to violence on their platform,” the groups said in their report.
On August 9, Kenya will go to the polls, which are expected to be tightly contested and bitterly fought.
Given Kenya’s recent history of electoral violence and the “polarised, ethnically driven and personalist politics” of the country, it remains vulnerable to unrest.
Some of the worst violence occurred after the 2007 elections, when tribal tensions were laid bare after inflammatory electoral campaigns and a disputed result. As many as 1,300 people were killed and hundreds of thousands fled their homes.
The investigation tested Facebook’s ability to detect hate speech ahead of the Kenyan elections.
The groups submitted 20 ads to Facebook, which covered the 10 real-life hate speech examples and their corresponding translation in English or Swahili.
“Much to our surprise and concern, all hate speech examples in both languages were approved, with one exception: our English language hate speech ads were initially rejected for failing to comply with Facebook’s Grammar and Profanity policy,” the report mentioned.
“Facebook invited us to update the ads, and after making minor corrections they were similarly accepted. Seemingly our English ads had woken up their AI systems, but not for the reason we expected,” it added.
A Meta spokesperson said that they’ve taken “extensive steps” to help Meta “catch hate speech and inflammatory content in Kenya” and that they’re “intensifying these efforts ahead of the election”.
The groups called on Facebook to urgently increase the content moderation capabilities and integrity systems deployed to mitigate risk before, during and after the upcoming Kenyan election.