The challenge
Abusive speech and imagery continually change. Negative stereotypes can be expressed indirectly, through coded language, humour, images and references that evade simple keyword-based moderation. Platforms therefore need community standards that reflect current scientific understanding of prejudice, stereotyping and the dynamics of online hate.
This is especially important for creators from minority communities, who may face targeted hostility that appears innocuous when individual posts are assessed without social or cultural context.
The approach
nisien.ai provided research-led consultancy to TikTok on its community guidelines. The work drew on expertise developed through HateLab in criminology, psychology, online behaviour and the computational detection of hate speech.
Rather than treating harmful content as a fixed vocabulary, the consultancy considered the psychological and behavioural processes that underpin online hate. These include negative stereotyping, unconscious prejudice, identity-based hostility and the way apparently minor expressions can contribute to a harmful environment when repeated or amplified.
The work translated academic evidence into guidance that could inform platform policy and creator safeguarding.
The outcome
TikTok’s updated community guidelines incorporated current scientific understanding of the psychological and behavioural foundations of online hate speech.
The project shows why online safety requires a combination of technical and social-scientific knowledge. Detection systems need clear, defensible definitions of harm, while policy teams need evidence about context, intent and impact. Bringing those perspectives together creates a stronger basis for protecting creators while respecting freedom of expression.