Dec 28, 2019 - Technology

Facebook purges accounts that used AI-generated profile photos

In this image, a man holds a smartphone and a digital line emitted from the phone covers his eyes.

Illustration: Aïda Amer/Axios

Hundreds of political Facebook accounts, groups and pages — and 72 Instagram accounts — have been purged after using fake, AI-generated profile photos to masquerade as Americans, the company announced.

Why it matters: This is the first time Facebook has seen “a systemic use" of AI-generated photos in profile pictures "to make accounts look more authentic," Facebook head of security policy Nathaniel Gleicher told the New York Times.

Yes, but: Gleicher also told the Times "that this AI technique did not actually make it harder for the company’s automated systems to detect the fakes, because the systems focus on patterns of behavior among accounts."

The big picture: Social media platforms face an uphill battle against misinformation ahead of the 2020 presidential election, as new tactics emerge and evolve.

Details: Many of the posts, which originated in the U.S. and Vietnam, targeted President Trump's impeachment and promoted conservative ideology and "family values and freedom of religion," Facebook said.

  • The accounts involved are tied to Epoch Media Group, parent company to the Epoch Times, Facebook found.
  • The group denied Facebook's findings in an email to the NYT.
  • The outlet was banned from placing ads on Facebook in August, after CNBC reported that it hid its connection to ads on the platform promoting Trump and conspiracy theories.

Go deeper: Misinformation haunts 2020 primaries

Go deeper