You obviously have the technology. If you're detecting 91% of the terrorism-related and violent extremist-related content, and you're able to detect them and remove them 91% of the time before they get on the platform, then you obviously have the technology to do a better job at preventing other types of online hatred, such as racist propaganda, threatening and abusive harassment. Why aren't you doing a better job with that type of conduct?
On May 30th, 2019. See this statement in context.