Our approach to misinformation

UPDATED

JUL 29, 2021

Facebook is committed to stopping the spread of misinformation. We use a combination of enforcement technology, human review and independent fact-checkers to identify, review and take action on this type of content.

Our strategy for misinformation: remove, reduce, inform

Remove

We value both free expression and keeping people safe, so we remove misinformation from Facebook technologies in limited cases:

  • When misinformation has the potential to cause imminent physical harm. For example, we remove false claims about COVID-19 and vaccines that leading health organizations have debunked. We do this to ensure safety.

  • When misinformation has the potential to interfere with or suppress voting. We do this because it undermines expression and voice.

  • When videos are manipulated in ways that would not be apparent to an average person—and would mislead an average person to believe a subject of the video said words they did not say—and were created using artificial intelligence that realistically combines, replaces or superimposes content onto them. We do this to ensure authenticity.

Taking down violating content

Reduce

When one of our fact-checking partners rates something as false, we reduce the content’s distribution in News Feed and other places so fewer people see it. When Pages and Groups repeatedly share such content, we reduce the distribution of all of their posts in News Feed and remove them from the recommendations we show people.

Reducing the distribution of problematic content

Inform

We apply notices to fact-checked posts and send notifications to the people who posted it. That way, people can see what the fact-checkers concluded and decide for themselves what to read, trust or share. We also partner with organizations around the world to promote news literacy.

Providing context on sensitive or misleading content

Our fact-checking program

Facebook’s fact-checking program launched in 2016 for the Facebook app and expanded to Instagram in 2019. It focuses on identifying and addressing viral misinformation, particularly hoaxes with no clear basis in fact.

Our fact-checking partners prioritize provably false claims, especially those that are timely, trending or important. They don’t prioritize claims that are inconsequential or contain only minor inaccuracies.

The program is also not intended to interfere with individual expression, opinions and debate, clearly satirical or humorous content or business disputes.

Content fact-checkers prioritize

The content fact-checkers prioritize

Content fact-checkers prioritize

UPDATED

JUL 29, 2021

Fact-checkers review and rate public Facebook posts and public Instagram posts, including ads, articles, photos, videos and text-only content. In the spirit of free expression, they don’t interfere with opinions and debate, content that’s clearly satirical or humorous or business disputes.

Prioritize

Information fact-checkers prioritize:

  • Viral false information.

  • Clear hoaxes that have no basis in fact.

  • Content presented as opinion, but based on underlying false information.

  • Provably false claims, especially ones that are timely, trending and important to the average person.

Don't prioritize

Information fact-checkers don’t prioritize:

  • Claims that are inconsequential.

  • Claims that consist of minor inaccuracies.

Not eligible to be fact-checked

Information that’s not eligible to be fact-checked:

  • Content that doesn’t include a verifiable claim.

  • Content that was true at the time of writing.

  • General opinion content, as long as it doesn’t spread false information. Note: Fact-checkers use their judgment to determine whether content is actually opinion or whether it is masking false information in the guise of opinion, and to rate it as appropriate in these circumstances.

  • Opinion and speech from politicians.

Prioritize
Don't prioritize
Not eligible to be fact-checked

Information fact-checkers prioritize:

  • Viral false information.

  • Clear hoaxes that have no basis in fact.

  • Content presented as opinion, but based on underlying false information.

  • Provably false claims, especially ones that are timely, trending and important to the average person.

Information fact-checkers don’t prioritize:

  • Claims that are inconsequential.

  • Claims that consist of minor inaccuracies.

Information that’s not eligible to be fact-checked:

  • Content that doesn’t include a verifiable claim.

  • Content that was true at the time of writing.

  • General opinion content, as long as it doesn’t spread false information. Note: Fact-checkers use their judgment to determine whether content is actually opinion or whether it is masking false information in the guise of opinion, and to rate it as appropriate in these circumstances.

  • Opinion and speech from politicians.

Fact-checkers

We work with more than 80 certified, independent fact-checking organizations in more than 60 languages around the world. These fact-checkers are certified through the non-partisan International Fact-Checking Network (IFCN), which is a subsidiary of the journalism research organization The Poynter Institute.

How fact-checking works

How fact-checking works

How fact-checking works

UPDATED

JUL 29, 2021

Although fact-checkers are independent from Facebook, we work together to address false information in our community. While fact-checkers focus on the legitimacy and accuracy of information, we focus on taking action by informing our community when content has been rated false. Here’s how it works.

Identifying false information

Our technology can detect posts that are likely to be misleading based on various signals — this includes how people are responding and how fast the content is spreading. Reports from people on Facebook and Instagram can also indicate that fact-checkers need to give a post a closer look. Other signals that help us identify false information include:

  • Comments on posts that express disbelief.

  • Feedback from people who report posts in News Feed as “false information.”

  • Machine learning models that continuously improve our ability to predict false information.

  • Fact-checkers identifying content to review on their own.

All fact-checkers in the International Fact-Checking Network (IFCN) have access to CrowdTangle—a public tool from Facebook that provides insights into how public content is performing on social media—to help them identify posts that contain misinformation. Publishers, journalists, researchers and academics also use CrowdTangle to follow, analyze and report on public content on social media.

Reviewing content

Fact-checkers will review a piece of content, check its facts and rate its accuracy. This process occurs independently from Facebook, and may include calling sources, consulting public data, authenticating images and videos and more.

Clearly labeling false information and informing people about it

When content has been reviewed by fact-checkers, we add a notice to it so people can read additional context. We also notify people before they try to share this content, including those who’ve shared it in the past.

Ensuring fewer people see false information

Once a fact-checker rates a piece of content as False, Altered Photo or Video or Partly False, it appears lower in Facebook News Feed, gets filtered out of the Explore tab on Instagram and is featured less prominently in feed and stories. This significantly reduces the number of people who see it. We may also reject ads with content that has been rated by fact-checkers.

Taking action against repeat offenders

Pages and websites that repeatedly share false information rated False or Altered Photo or Video will have some restrictions. This includes having their distribution reduced, their ability to monetize and advertise removed and their ability to register as a news Page removed for a given time period.

Content ratings fact-checkers use in their work

Content ratings fact-checkers use in their work

UPDATED

JUL 29, 2021

Fact-checkers use content ratings to help people better judge a piece of content and decide whether to read, trust or share it. Here are some of the ratings fact-checkers use in their work.

False

False

A False rating is for content that has no basis in fact. This includes:

  • Fake quotes.

  • Impossible claims, or ones that can’t be considered an interpretation of something that actually happened or was said. This could be claiming a natural disaster took place when no such event happened.

  • Fabricated content from websites misrepresenting themselves as real news outlets.

  • Image, audio or video content that’s authentic but offered as proof of an unrelated event.

Content rated False will appear lower in News Feed, be filtered out of Explore on Instagram, and be featured less prominently in feed and stories. This significantly reduces the number of people who see it.

Partly False

Information fact-checkers don’t prioritize:

A Partly False rating is for content that has some factual inaccuracies. This includes:

  • Inaccuracies or miscalculations regarding numbers, dates and times, but that could be considered an interpretation of something that actually happened or was said. This could be miscalculating the cost of a government program.

  • A mix of true and false information.

  • Content presented as an opinion but based on underlying false information.

Altered Photo or Video

Information that’s not eligible to be fact-checked:

An Altered rating is for images, audio and videos that’ve been manipulated in ways that could mislead people. This includes:

  • An edited video showing a person shaking someone’s hand when they didn’t.

  • An image where someone used Photoshop to depict a person at a location they weren’t actually at.

  • A media outlet editing to omit or reorder the words someone said to reverse the meaning of the statement. This could be removing the word “not” from someone saying “I will not do [x].”

Content rated Altered will appear lower in News Feed, be filtered out of Explore on Instagram, and be featured less prominently in feed and stories. This significantly reduces the number of people who see it.

Missing Context

Missing Context

A Missing Context rating is for content that might mislead people without additional context. This includes:

  • Clips from authentic video or audio, or cropping of authentic photos, that lacks the full context from the original content, but that haven’t otherwise been edited or manipulated. This could be cropping a video clip of someone saying “I support that candidate if...” to only include “I support that candidate.”

  • Media content edited to omit or reorder the words someone said that changes, but that does not reverse, the meaning of the statement.

  • Hyperbole or exaggeration that is technically false but based on a real event or statement.

  • Content that presents a conclusion not supported by the underlying facts.

  • Claims stated as fact that are unproven.

Satire

Satire

A Satire rating is for content that uses irony, exaggeration or absurdity for criticism or awareness — especially in the context of political, religious or social issues — but that a reasonable person wouldn’t immediately understand to be satirical. This includes:

  • Content from sites not clearly labeled as or widely-known as satire.

  • Content presented without clear labeling.

When a fact-checker rates a post as Satire, an article will be appended to the bottom of the post, but it won’t be labeled Satire.

False
Partly False
Altered Photo or Video
Missing Context
Satire

False

A False rating is for content that has no basis in fact. This includes:

  • Fake quotes.

  • Impossible claims, or ones that can’t be considered an interpretation of something that actually happened or was said. This could be claiming a natural disaster took place when no such event happened.

  • Fabricated content from websites misrepresenting themselves as real news outlets.

  • Image, audio or video content that’s authentic but offered as proof of an unrelated event.

Content rated False will appear lower in News Feed, be filtered out of Explore on Instagram, and be featured less prominently in feed and stories. This significantly reduces the number of people who see it.

Information fact-checkers don’t prioritize:

A Partly False rating is for content that has some factual inaccuracies. This includes:

  • Inaccuracies or miscalculations regarding numbers, dates and times, but that could be considered an interpretation of something that actually happened or was said. This could be miscalculating the cost of a government program.

  • A mix of true and false information.

  • Content presented as an opinion but based on underlying false information.

Information that’s not eligible to be fact-checked:

An Altered rating is for images, audio and videos that’ve been manipulated in ways that could mislead people. This includes:

  • An edited video showing a person shaking someone’s hand when they didn’t.

  • An image where someone used Photoshop to depict a person at a location they weren’t actually at.

  • A media outlet editing to omit or reorder the words someone said to reverse the meaning of the statement. This could be removing the word “not” from someone saying “I will not do [x].”

Content rated Altered will appear lower in News Feed, be filtered out of Explore on Instagram, and be featured less prominently in feed and stories. This significantly reduces the number of people who see it.

Missing Context

A Missing Context rating is for content that might mislead people without additional context. This includes:

  • Clips from authentic video or audio, or cropping of authentic photos, that lacks the full context from the original content, but that haven’t otherwise been edited or manipulated. This could be cropping a video clip of someone saying “I support that candidate if...” to only include “I support that candidate.”

  • Media content edited to omit or reorder the words someone said that changes, but that does not reverse, the meaning of the statement.

  • Hyperbole or exaggeration that is technically false but based on a real event or statement.

  • Content that presents a conclusion not supported by the underlying facts.

  • Claims stated as fact that are unproven.

Satire

A Satire rating is for content that uses irony, exaggeration or absurdity for criticism or awareness — especially in the context of political, religious or social issues — but that a reasonable person wouldn’t immediately understand to be satirical. This includes:

  • Content from sites not clearly labeled as or widely-known as satire.

  • Content presented without clear labeling.

When a fact-checker rates a post as Satire, an article will be appended to the bottom of the post, but it won’t be labeled Satire.