Prepare for lies: Facebook has finally turned off the AI ​​system that detected viral fakes

Facebook is one of the best places to spread propaganda. It has been chosen by Russia, China, Iran, various scammers with generated images, etc. The company Meta, which operates the social network, has created a machine learning system to detect and remove at least some of these fakes. However, Mark Zuckerberg has now announced a change in fact-checking policies amid Donald Trump's rise to power.

The Age of Disinformation

According to recent reports, after the recent announcement the company has already disabled its viral misinformation tracking systems, which targeted crazy and sometimes physically dangerous narratives, conspiracy theories, propaganda and fake news. Interestingly, Meta itself previously admitted that these efforts were extremely effective: machine learning tools limited the spread of viral disinformation with an impressive efficiency of over 90 percent. However, in his video address, the head of the company still stated that this approach is imperfect, and therefore the company is switching to so-called community notes, where users verify the information themselves, reports 24 Kanal with reference to Platformer.

Beyond the video Zuckerberg posted, Meta has not publicly commented on the decision to disable its misinformation trackers. But last week, the company’s newly promoted head of global policy said in a blog post that Meta’s safety practices unfairly downgraded “too much content” that the systems predicted might violate the rules but actually did not. He did not specify what that content was.

We are in the process of removing most of these downgrades and requiring greater assurance that content violates the standards for the rest,
– says the post, which was made before the final shutdown of tracking systems.

Zuckerberg, in his lengthy interview with Joe Rogan, complained about censorship by the Biden administration (although it existed before he took office). And so these changes are now being presented as a return to free speech, or, as he put it in his address, “a return to my roots of freedom of expression.”

These supposedly free speech-oriented changes include not only the closure of the fact-checking program, but also the granting of permission for offensive language: users of all Meta platforms are now allowed to insult transgender people, call women “household items or property,” declare themselves “proud racists,” and more.

It is currently unclear how these numerous changes will ultimately affect the daily experience of Meta users, let alone the broader information ecosystem. But they certainly appeased the new president, Donald Trump, who had previously threatened to imprison Mark Zuckerberg over false accusations of election interference.