The service said it had conducted a review of the videos and said there was language that was “clearly hurtful,” but that there hadn’t been policy violations. In a response to Gizmodo, a Google spokesperson also claimed that Crowder was only “debating” Maza’s opinions, and was off the hook as he hadn’t directly “instructed” viewers to harass the Vox host. Crowder posted an apology video (though some have questioned its sincerity), but simultaneously claimed he hadn’t violated policies.

There are “other aspects” of Crowder’s channel that YouTube looking at, the company added.

The problem, as you might imagine, is that Crowder’s statements appear to explicitly violate YouTube’s policies on hate speech and harassment. The company forbids slurs and stereotypes that “incite or promote hatred,” and its anti-harassment policy bars any content that makes “hurtful and negative personal comments/videos” about other people. Crowder was launching unambiguous personal attacks against Maza, and it’s hard to imagine him being unaware of the likely consequences when he has millions of viewers.

The incident underscores a recurring issue with internet companies’ frequently changing approaches to hate speech. While they will take down hateful content, they may only use a narrow definition of hate to dictate their responses, regardless of their policies — and even then, they might only take action after outside pressure forces their hand. There isn’t always a proactive attempt to curb hate, and people like Crowder can slip through the cracks as a result.