BranchVoice
Demo

Hate Speech Detector: Filter Harmful Content Automatically

Shows how likely the text is to be harmful and of which kind. It tries to tell harsh but legitimate criticism from abuse.

0 / 1500 characters

Examples

Result

The result will appear here.

Results are AI estimates and may contain errors. Your text is not stored; the result for the same text is cached briefly.

Labels

The model scores the probability of each label below for every text. You can extend the label set for your needs or use it with labels made entirely for you.

Hate speech
Attacks on ethnicity, religion, gender, sexual orientation, disability or nationality.
Insult
Insulting or belittling a person or group.
Profanity
Swearing and vulgar language.
Threat
Threats of violence or harm.
Harassment
Harassment or bullying aimed at a specific person.
Discrimination
Calls to exclude a group or treat it unequally.

Where is it used?

Examples of how teams use this model in their daily work.

  • 01

    Comment and community moderation

    Screen user comments on your site, app or forum automatically before they go live.

  • 02

    Protecting social accounts

    Filter insults and threats on your brand accounts so your team focuses on real customer messages first.

  • 03

    Employee safety

    Detect threats and harassment aimed at staff and pass them to HR and security.

  • 04

    Brand safety

    Check content for harmful language before running ads or sponsorships next to it.

Use this model in your own system

Change the labels, add new ones for your industry, or connect the model to your CRM, contact centre or e-commerce stack through our API. For API access and custom models, write to us at [email protected].

Write to us

Frequently asked questions

Is this tool free?
Yes. You can analyse a number of texts a day without signing in; heavy use gets a short cool-down.
Which languages does it support?
It understands multilingual text, Turkish and English first; results are shown in your interface language.
Is my text stored?
No. The text is only processed for the analysis; the result is cached briefly so repeated requests return quickly.
Can I analyse in bulk?
The BranchVoice platform collects your reviews from their sources automatically and classifies them with these models in bulk; request a demo to try it.

Other tools

All tools

Want to analyse thousands of reviews automatically?

BranchVoice collects reviews from Google Maps, complaint sites, social media, app stores and marketplaces every day, classifies them with these models and turns them into alerts, reports and dashboards.

Request a demo