Meta
Meta
Policies
Community StandardsMeta Advertising StandardsOther policiesHow Meta improvesAge-appropriate content
Features
Our approach to dangerous organisations and individualsOur approach to the opioid epidemicOur approach to electionsOur approach to misinformationOur approach to newsworthy contentOur approach to Facebook Feed rankingOur approach to explaining rankingAccessibility at Meta
Research tools
Content Library and Content Library APIAd Library toolsOther research tools and datasets
Enforcement
Detecting violationsTaking action
Governance
Governance innovationOversight Board overviewHow to appeal to the Oversight BoardOversight Board casesOversight Board recommendationsCreating the Oversight BoardOversight Board: Further asked questionsMeta's biannual updates on the Oversight BoardTracking the Oversight Board's impact
Security
Threat disruptionsSecurity threatsThreat reporting
Reports
Community Standards enforcement reportIntellectual propertyGovernment requests for user dataContent restrictions based on local lawInternet disruptionsWidely viewed content reportRegulatory and other transparency reports
Policies
Community Standards
Meta Advertising Standards
Other policies
How Meta improves
Age-appropriate content
Features
Our approach to dangerous organisations and individuals
Our approach to the opioid epidemic
Our approach to elections
Our approach to misinformation
Our approach to newsworthy content
Our approach to Facebook Feed ranking
Our approach to explaining ranking
Accessibility at Meta
Research tools
Content Library and Content Library API
Ad Library tools
Other research tools and datasets
Enforcement
Detecting violations
Taking action
Governance
Governance innovation
Oversight Board overview
How to appeal to the Oversight Board
Oversight Board cases
Oversight Board recommendations
Creating the Oversight Board
Oversight Board: Further asked questions
Meta's biannual updates on the Oversight Board
Tracking the Oversight Board's impact
Security
Threat disruptions
Security threats
Threat reporting
Reports
Community Standards enforcement report
Intellectual property
Government requests for user data
Content restrictions based on local law
Internet disruptions
Widely viewed content report
Regulatory and other transparency reports
Policies
Community Standards
Meta Advertising Standards
Other policies
How Meta improves
Age-appropriate content
Features
Our approach to dangerous organisations and individuals
Our approach to the opioid epidemic
Our approach to elections
Our approach to misinformation
Our approach to newsworthy content
Our approach to Facebook Feed ranking
Our approach to explaining ranking
Accessibility at Meta
Research tools
Content Library and Content Library API
Ad Library tools
Other research tools and datasets
Security
Threat disruptions
Security threats
Threat reporting
Reports
Community Standards enforcement report
Intellectual property
Government requests for user data
Content restrictions based on local law
Internet disruptions
Widely viewed content report
Regulatory and other transparency reports
Enforcement
Detecting violations
Taking action
Governance
Governance innovation
Oversight Board overview
How to appeal to the Oversight Board
Oversight Board cases
Oversight Board recommendations
Creating the Oversight Board
Oversight Board: Further asked questions
Meta's biannual updates on the Oversight Board
Tracking the Oversight Board's impact
Policies
Community Standards
Meta Advertising Standards
Other policies
How Meta improves
Age-appropriate content
Features
Our approach to dangerous organisations and individuals
Our approach to the opioid epidemic
Our approach to elections
Our approach to misinformation
Our approach to newsworthy content
Our approach to Facebook Feed ranking
Our approach to explaining ranking
Accessibility at Meta
Research tools
Content Library and Content Library API
Ad Library tools
Other research tools and datasets
Enforcement
Detecting violations
Taking action
Governance
Governance innovation
Oversight Board overview
How to appeal to the Oversight Board
Oversight Board cases
Oversight Board recommendations
Creating the Oversight Board
Oversight Board: Further asked questions
Meta's biannual updates on the Oversight Board
Tracking the Oversight Board's impact
Security
Threat disruptions
Security threats
Threat reporting
Reports
Community Standards enforcement report
Intellectual property
Government requests for user data
Content restrictions based on local law
Internet disruptions
Widely viewed content report
Regulatory and other transparency reports
English (UK)
Privacy PolicyTerms of ServiceCookies
Home
Enforcement
Detecting violations
Challenge of detecting hate speech

The challenge of detecting hate speech

UPDATED 19 JAN 2022
Hate speech is especially difficult for technology and human review teams to detect. Idioms and nuances vary widely across cultures, languages and regions. Also, people sometimes share words that would normally be hate speech, but they do it to raise awareness for the problem or to use self-referentially in an effort to reclaim the term.
Those are challenges just detecting hate speech in text. A lot of hate speech we find on Facebook and Instagram is in photos or videos. A meme, for example, might use text and images together to attack a particular group of people. This is an even greater challenge for technology.
hate-speech-image
Content like this gets more complicated when people try to avoid detection by changing their content. For example, they might misspell words, avoid certain phrases or modify their images and videos.
Progress in using artificial intelligence to detect hate speech
We improved our tools for detecting hate speech over the last several years, so now we remove much of this content before people report it – and, in some cases, before anyone sees it.
We use AI to identify images and text that are identical to content that we have already removed as hate speech. Our technology also looks at reactions and comments to assess how similar a piece of content is.
These techniques help our technology more accurately detect hate speech, even when the meaning is not obvious or the content is changed to avoid detection.
Read an in-depth analysis on the Facebook AI blog