Kick’s 100M-User Platform Admits Hate Speech Detection Is ‘Art Not Science’

Hate speech is no longer just a policy violation—it’s an early warning system for digital threats. Kick’s admission before Australia’s Royal Commission that it can’t reliably detect anti-Semitic slurs on a platform of over 100 million users underscores a systemic cybersecurity failure. When moderation becomes a guessing game outsourced to a distant team, the door opens for coordinated disinformation, extremist recruitment, and real-world violence that threat intelligence teams cannot afford to ignore.

A shocking admission from a top executive at live-streaming platform Kick has exposed critical vulnerabilities in the detection of online hate speech, underscoring the broader systemic failures across major social media networks. Appearing before Australia’s Royal Commission into Anti-Semitism on July 6, 2026, Kick’s General Counsel Tiat Oon Ooi was unable to definitively confirm that calling Jews “evil rats and subhumans” would violate the platform’s community guidelines. His testimony, which described content moderation as “more an art than a science,” reveals the profound ambiguity that allows hate speech to persist at scale on digital platforms with over 100 million users.

“

Appearing before Australia’s Royal Commission into Anti-Semitism on July 6, 2026, Kick’s General Counsel Tiat Oon Ooi was unable to definitively confirm that calling Jews “evil rats and subhumans” would violate the platform’s community guidelines.

Kick, Melbourne-headquartered and self-styled as a higher-paying, looser-moderation alternative to Amazon’s Twitch, has rapidly grown its user base precisely by streamlining moderation. Yet this approach now faces intense scrutiny. Ooi conceded that only the most egregious violations—such as explicit calls to violence—are clear-cut, while nearly everything else falls into a gray zone where his company’s moderation policies fail. He admitted not even being “specifically well-acquainted with the details of the moderation guidelines,” a startling disclosure from a general counsel overseeing a platform of such scale.

The revelation is compounded by Kick’s reliance on an outsourced content moderation team in Serbia. Geo-political and linguistic distance from the platform’s predominantly English-speaking user base inevitably introduces delays and cultural misinterpretation. Ooi attributed the platform’s lower removal rates for reported posts to what he called “a significant number of false alarms” triggered by users who simply dislike a streamer. However, this defense highlights a dangerous reliance on reactive, report-based moderation rather than proactive, AI-driven detection—a gap that threat actors can easily exploit to propagate extremist narratives.

Meanwhile, Meta’s earlier testimony to the same inquiry exposed a different but equally troubling philosophy. The parent company of Facebook, Instagram, and Threads acknowledged it treats anti-Semitic hate speech as categorically less severe than “truly heinous” content like glorification of terrorism. This tiered approach effectively creates a permitted zone for certain types of hate speech, allowing corrosive ideologies to fester and often serve as precursors to violent radicalization. For cybersecurity and threat intelligence professionals, this is a glaring red flag: hate speech is not merely offensive content but often a leading indicator of coordinated information operations, recruitment for extremist groups, and eventual offline violence.