> With sensitivity settings, brands can mark their accounts as "conservative"—avoiding ad placements next to "targeted hate speech, explicit sexual content, gratuitous gore, excessive profanity, obscenity, spam, drugs"—or "standard"—only avoiding placements next to "targeted hate speech, explicit sexual content, gratuitous gore, excessive profanity." Soon there will be a third tier, "relaxed," where brands can "show ads alongside some sensitive content to maximize reach" but still cannot monetize targeted hate speech or explicit sexual content.

Interesting, it sounds like they're able to reliably detect "targeted hate speech, explicit sexual content, gratuitous gore, excessive profanity, obscenity, spam, drugs"