Skip to main content

Twitter’s upcoming manipulated media policy may include labels for ‘harmfully misleading’ tweets

Twitter’s upcoming manipulated media policy may include labels for ‘harmfully misleading’ tweets


A project under consideration includes colored labels for tweets that contain lies

Share this story

The Twitter bird logo in white against a dark background with outlined logos around it and red circles rippling out from it.
Illustration by Alex Castro / The Verge

A leaked Twitter demo showed that the platform is considering labeling tweets from politicians and other public figures if the tweets have lies or misinformation in them, NBC News reports. “Harmfully misleading” tweets would have red or orange labels applied, and include corrected information from fact-checkers, journalists, and other users in “one possible iteration” of a Wikipedia-like anti-misinformation effort Twitter plans to unveil March 5th, according to NBC News.

The project is in the very early stages at Twitter, but a Twitter spokesperson says in an email to The Verge that the labeling system is a “design mockup for one option that would involve community feedback. Misinformation is a critical issue and we will be testing many different ways to address it.” 

The leaked demo showed a tweet from House Minority Leader Kevin McCarthy (R-CA) that had incorrect information about whistleblowers and one from Sen. Bernie Sanders (I-VT) that had an incorrect figure about background checks on gun sales, NBC News reports, as well as a tweet with false information about the coronavirus. It’s not clear, however, how Twitter would determine which tweets would be flagged with the new labels.

Image: Twitter via NBC News

One version of the leaked demo — which the company stressed to NBC News was just one of several possibilities — would allow Twitter users / moderators to earn points or “community badges” if they “contribute in good faith and act like a good leader.” How it would choose which users get to be moderators and what kind of authority moderators would have is unclear, but “the more points you earn, the more your vote counts,” according to the demo. “Community” members would be asked to rate a tweet as “likely” or “unlikely” to be “harmfully misleading”; to predict how many other community members would give the same answer using a sliding scale of 1 to 100; and to elaborate on why they believe the tweet should be flagged.

Earlier this month, Twitter announced the March 5th rollout of its new “manipulated media” policy which will include a ban on fabricated photos, videos, and media on the platform that is “deceptively shared,” and may pose safety risks. Media that has been deceptively edited or otherwise altered in a way that changes its meaning would be labeled as fake, the company said.

Twitter has put other tools into place in an attempt to help users discern what information on its platform is inaccurate. In May, it announced a tool that could tamp down anti-vaccination information by redirecting users to, a site run by the Health and Human Services Department. And last week, Twitter announced it was expanding its election integrity policy, which prohibits users from sharing “false or misleading information about how to participate in an election or other civic event” by redirecting users to when they search for certain keywords associated with the 2020 US census.