Tool: Content Moderation & Toxicity Scorer

Score any text for toxicity and unsafe content in one run. Get a clear safe flag, an overall toxicity score and per category scores for hate speech, harassment, sexual, violent, self harm and spam content.

📥 Input · 📤 Output · 💰 Pricing · ▶️ Examples

Safe flag true or false per text

7 categories scored 0 to 1

JSON / CSV / Excel output formats

What it does

Give it a list of texts. For each one it returns a clean, structured moderation record:

• safe: a single true or false decision
• overallToxicity: a score from 0 to 1
• categories: individual scores from 0 to 1 for toxicity, hate speech, harassment, sexual, violence, self harm and spam
• flagged: the categories that crossed your threshold
• reason: a short plain explanation of the main risk

Use it to screen user comments, chat messages, reviews, form submissions or any user generated content before it goes live. Export to JSON, CSV or Excel, or pull it through the Apify API.

Paid plans only

This tool runs on the latest, most reliable premium AI models. Because each run has a real model cost, it is available to paid Apify accounts only. A run started from a free account stops immediately with an upgrade message. Upgrade your Apify subscription to a paid plan and run it again.

Who is it for

Use case	Who benefits
Community platforms	Teams moderating comments and posts at scale
Marketplaces	Screening reviews and listings before publish
Support desks	Flagging abusive or unsafe inbound messages
Developers	Adding a moderation step to any content pipeline

Input

• texts: a list of items, each a plain string or an object with a text field.
• threshold: the score at which a category is flagged (0 to 1). Optional.
• maxItems: optional cap on how many texts to score this run.

Notes

• Scores are model estimates, not legal judgements. Set your threshold to match your own policy and review borderline cases.
• Long texts are truncated before scoring to keep runs fast and predictable.

Example use cases

Ready-to-run example tasks, each preconfigured for a common scenario. Open one and press run, or use it as a template:

• Moderate a friendly support message: Score a customer support message for toxicity and safety flags and return the results as JSON for your moderation queue.
• Moderate a suspected spam message: Detect spam and promotional content in a message and get toxicity and safety scores as JSON for filtering user posts.
• Moderate a harassment comment: Flag harassment and abusive language in user text and return category scores as JSON for community safety systems.
• Moderate a product review for safety: Check product reviews for offensive or unsafe content and return moderation scores as JSON before publishing them.
• Moderate a public forum post: Analyze a forum post for hate, harassment and unsafe content and return JSON scores for automated moderation rules.
• Moderate a neutral scheduling notice: Run a routine message through content moderation to confirm it is safe and return category scores as clean JSON.
• Moderate a batch of user comments: Score several user comments at once for toxicity and safety and return per text results as JSON for bulk moderation.
• Moderate comments with a strict threshold: Apply a low threshold to catch borderline unsafe comments and return detailed category scores as JSON for strict review.
• Moderate comments with a lenient threshold: Use a high threshold to flag only clearly harmful content and return moderation scores as JSON for light filtering.
• Moderate a mix of safe and toxic texts: Screen a mix of safe and toxic texts in one run and return toxicity and safety scores as JSON for your review pipeline.

Export, API and AI agents (x402 + MCP)

Export the scraped data to JSON, CSV or Excel, pull it as a dataset through the Apify API, or wire it into your app with no code. This web scraper and data extractor also works for bulk data extraction and scheduled runs.

For AI agents: this Actor is available on x402, Apify's agentic payment standard built with Coinbase. An AI agent can discover, pay for and run it on its own with a funded wallet and a single HTTP request: no account, no subscription, no API key and no human in the loop. It also runs as an MCP tool inside Claude, Cursor and other AI clients out of the box. Learn more about x402 agentic payments on Apify.

Built by scrapers.lat.