Moderation rules

Introduction

This manual defines elv.ai’s comprehensive approach to content moderation with transparency. Our objective is to ensure a secure and positive online environment by systematically eliminating profanity, hate speech, fraud, misinformation, and hoaxes, each of which is meticulously defined in the respective sections of this manual. Our guidelines are formulated in alignment with the prevailing Meta Community Guidelines and the provisions of the Digital Services Act legislation.

Categories of harmful content

Vulgarisms

Vulgarisms encompass comments containing obscene, profane, or offensive language and expressions, including crude vulgarisms, abbreviations of vulgarisms, and “censored” forms of these words.

Hate speech

Hate speech includes comments that insult or threaten individuals or specific groups, or incite other users to hate those individuals or groups through their content.

Fraud (Scams)

Scams involve attempts to deceive users and obtain sensitive information or funds from them through false profiles or other fraudulent practices.

Disinformation and hoaxes

Disinformation and hoaxes are false or misleading information disseminated to negatively influence public opinion. This includes unsubstantiated information about political and public events, public figures and the like. Example. WELL WE ALL KNOW THAT COVID IS AN INVENTED ‘VIRUS’!!!”

Conclusion

These rules are binding for moderators and elv.ai’s AI model is trained based on them. The rules are fixed, though additional rules may be added to address brand-specific expressions based on special client needs.