Product1 publisher2 min readPublished
Meta recruits Community Notes writers in 16 Latin American countries before any note goes live
Contributors are being onboarded in Argentina, Mexico, Venezuela and 13 other countries, and Meta's own blog post says a full launch would come with a transition away from its paid third-party fact checkers.
The Product Desk · Product desk
What happened
- Meta has begun testing Community Notes in 16 Latin American countries, more than a year after it launched the program in the US and ended its partnerships with US-based fact checking organizations.
- Approved contributors can write notes during the test, but Meta says those notes will not be surfaced to users and that third-party fact checking is unchanged for now.
- Indicator found that 3 percent of Instagram posts debunked by the AFP's US bureau since last April had also received a Community Note, against Meta's 1.4 million US contributors.
Compiled by The Product DeskSomething wrong?How this is made
Why it matters
- decision Any organisation holding a Meta fact-checking contract in these 16 markets now has to plan headcount against a transition the company has put in writing and has not dated.
- constraint With notes hidden for the duration of the test, no partner or regulator can measure the crowd layer's coverage in these markets before the paid layer is wound down.
- contradiction The blog post points at transitioning third-party fact checking while the spokesperson says nothing changes at this time, and which one a partner plans against decides whether it staffs for continuity or for wind-down.
- precedent Meta ran the US sequence of launching notes and then ending paid contracts, and the Oversight Board's warning about polarised and conflict-affected countries did not keep those countries off the second region's list.
A contributor approved in Colombia this month can write notes on Facebook, Instagram and Threads that nobody outside Meta reads [2][3][4]. The paid checkers keep their queue, and the volunteer's notes stay unpublished. The same opacity applies to the coverage data: a fact-checking organisation in one of these markets cannot see what share of its own debunks the crowd layer also caught.
One outside measurement exists, and it comes from the market Meta calls a success [11]. The publication Indicator took Instagram posts that the AFP's US bureau debunked since last April and found 3 percent had also received a Community Note [12]. That leaves 97 percent of the posts a professional bureau judged worth debunking with no note attached [1]. Meta's headline figure for the same program is 1.4 million contributors [11]. The contributor count measures how many people signed up, not what share of debunked posts got a note.
The Oversight Board's policy advisory opinion earlier this year ran 15,000 words [8]. It cautioned against bringing Community Notes to countries experiencing "high polarization," "protracted conflict" and those with a history of organized disinformation networks [9]. The International Fact-Checking Network says the test list includes exactly those cases, citing Nicaragua, which it says "has abolished elections and is actively practicing arbitrary detention of political dissidents," documented disinformation networks in El Salvador and Ecuador, and Venezuela's extended instability [10]. That is four of the sixteen test countries, a quarter of the list [2].
"One doesn't have to replace or substitute for the other, they can coexist," board co-chair Paolo Carozza told Engadget when the opinion was published [14]. The IFCN's statement, published by Poynter, said "the community notes program alone is simply not adequate when it comes to reducing the impact of harmful falsehoods on its platforms" [6].
Meta's blog post says it will "look at fully launching" once it is confident the product works, and describes that step as "beginning to publish Community Notes and transitioning our third-party fact checking program" [5]. A Meta spokesperson said there were "no changes to our third-party fact checking program at this time" [13]. Neither statement gives a date.
Anyone running a paid review layer next to a volunteer one can run the comparison Indicator ran, in-house, this quarter. It is a question of what the paid layer flagged, what share of that the volunteer layer also flagged, and what overlap rate would be acceptable before cutting the contract. The US pool is 1.4 million and the measured overlap on AFP's Instagram set is 3 percent [11][12].
What to watch
- The first published Community Note in any of the 16 countries, which is the signal the test has converted to a launch.
- Whether Meta gives its Latin American fact-checking partners notice periods or renewal terms, and how long they run.
- Any coverage measurement in the test markets comparing notes written to posts fact checkers debunked.