TRUST & SAFETY TAXONOMY
210 detection labels.
Not one generic moderation score.
Sentiqra's Trust & Safety taxonomy covers 21 policy areas and 210 detection labels. Detection labels are independently scored, giving your policy engine more precise signals for routing and enforcement. Configure different actions for different behaviors rather than treating every harmful signal the same.
5,000 analyzer runs/month · No credit card required
21 POLICY AREAS · 210 DETECTION LABELS
REPRESENTATIVE SIGNALS
profanityProfanity
insultInsult
harassmentHarassment
bullyingBullying
hate_speechHate Speech
race_based_hostilityRace-Based Hostility
dehumanizationDehumanization
threatThreat
incitement_to_violenceIncitement To Violence
sexual_harassmentSexual Harassment
The complete 210-label taxonomy is available to customers. Request access to get started.
No detection labels found. Try a different search term.
HARM GROUPS
Explore each harm group
Click any group to see its detection labels. Representative labels are shown publicly - the complete taxonomy is available to customers.
Profanity & Obscenity
Profanity, vulgarity, and obscene language directed at individuals or used aggressively.
Insults & Derogation
Direct insults, name-calling, belittling, and derogatory labeling of individuals.
Harassment & Targeting
Sustained harassment, repeated targeting, and persistent abuse of individuals.
Bullying & Social Exclusion
Bullying, cyberbullying, social exclusion, and coordinated harassment campaigns.
Hate & Identity Attack
Hate speech, identity-based attacks, and slurs targeting identity groups.
Identity-Based Discrimination
Discrimination and hostility based on race, ethnicity, religion, gender, and other identities.
Dehumanization & Exclusionary Harm
Dehumanizing language, vermin comparisons, contamination rhetoric, and extermination advocacy.
Threats & Intimidation
Direct and veiled threats of violence, harm, property damage, and coercion.
Incitement & Glorification of Violence
Inciting, encouraging, or celebrating violence, assault, and killing.
Sexual Harassment & Sexualized Abuse
Sexual harassment, unwanted sexual comments, sexualization, and sexual threats.
Coercive & Controlling Abuse
Coercion, blackmail, extortion, emotional abuse, and controlling behavior.
Self-Harm & Suicide Encouragement
Encouraging, glorifying, or pressuring others toward self-harm and suicide.
Self-Harm & Suicidal Ideation
First-person disclosure of suicidal thoughts, self-harm intent, or suicide planning.
Privacy Abuse & Doxxing
Doxxing, exposing personal information, and privacy invasion threats.
Scams, Fraud & Impersonation
Scams, phishing, impersonation, and fraudulent solicitations.
Spam & Manipulative Platform Abuse
Spam, engagement manipulation, and platform abuse tactics.
Child Safety & Predatory Language
Predatory language, grooming, and sexualization targeting minors.
Violent Extremism & Recruitment
Violent extremist propaganda, recruitment, and glorification of terrorism.
Dangerous Behavior & Harm Encouragement
Encouraging dangerous challenges, reckless behavior, and harmful dares.
Toxic Conflict & Hostile Conversation
Toxic language, hostile tone, inflammatory remarks, and flame war escalation.
Evasion & Coded Abuse
Masked profanity, coded hate, obfuscated threats, and dogwhistle language.
READING THE OUTPUT
What a Trust & Safety result tells you - and what it doesn't.
Intended task
Tell your moderation pipeline which documented harm categories a piece of text matches - the policy routing decision stays with you.
Input requirements
text- the only field analyzed; up to 4,096 Unicode code points. Over the limit is rejected withinput_too_long, never truncated.language- must be a language this analyzer supports. Unsupported codes are rejected with400 unsupported_language, never analyzed as a different language.analyzers- include"trust-safety"; up to 2 analyzers per request, one billed run each.- No context beyond
text- no user IDs, history, profiles, or attachments are read or inferred. - Works on the submitted text alone - the analyzer never sees the author's account, history, or reputation.
Ordinary positive
“You are an idiot and everyone on this server hates you. Keep talking and I will post your home address.”
Safe inference: The text matched two documented harm categories. You can route it to your harassment/doxxing policy lane and cite the exact labels in a statement of reasons.
Confusing negative · Quotation
“The commenter told her "I will post your home address" - our report on doxxing in gaming forums.”
What happens: A privacy/doxxing signal can still fire. The analyzer scores the submitted text as written - quoted abuse language inside reporting or commentary can match the same documented pattern.
Why it's tricky: A signal means "this text contains language matching the documented pattern," not "the author intended harm." Whether quotation context matters is a policy decision - keep human review in the loop for reported-speech surfaces.
What you can safely infer
- Which documented indicators the text matched, and at what signal-match strength.
- That the same input on the same pinned configuration replays identically - the receipt records which configuration produced the result.
- That a category-level signal with label-level children gives you both the policy area and the specific behavior.
What you must not infer
- ✕That the author intended harm - intent is not in the text.
- ✕That quoted or reported abuse means the author endorses it.
- ✕That an empty signals list clears the content - "not detected" is never "safe."
- ✕Anything about the author's account or history - the analyzer never sees it.
Features
210 indicators, 16 languages, sub-500ms p95 target
Use Cases
Social, gaming, marketplace, education
Playground
Test detection live - no API key needed
Benchmark Results
Per-language Top-3 Hit Rate benchmarks
Quickstart Guide
First API call in under 5 minutes
API Reference
Full endpoint, parameter, and response docs
vs LLM Moderation
Sentiqra vs Perspective, OpenAI, GPT-4o
Pricing
From Free to Enterprise - 30M calls/month
Detect all 210 indicators in real-time.
Every indicator is independently scored with confidence. 5,000 analyzer runs/month. No credit card required.