leading to what they viewed as excessive censorship of legitimate speech. It estimated that 10-20% of content removals were potentially mistakes. The platform plans to reduce restrictions on topics like immigration and gender identity and limit automated content scanning to serious violations such as ...