A logo of US company's Meta is displayed during the Vivatech technology startups and innovation fair, at the Porte de Versailles exhibition center in Paris, on May 22, 2024. Julien de Rosa, AFP/file

MANILA — Meta has been “working on responding better” against the spread of deepfake videos and other disinformation by developing more tools to police manipulated content that is “difficult” to detect, an official from the tech giant said on Monday.

Meta will soon “come up with tools and platforms that are useful for researchers” but the effort is “something underway and it is something that is [still] being designed," said Manu Gummi, Meta’s Public Policy Manager for Content.

“It is difficult for us to identify these but we are getting better at it… We are working on responding better. We’ve have more people working on this,” Gummi told journalists at this year’s International Media Conference.

“Our goal is to prevent election interference on Meta, and make it easier for people to have their voices heard in the political process… to make sure that we are responding with agility and accuracy,” she said.

Gummi did not disclose at what rate Meta increased its spending for its efforts vs disinformation.

In 2022, Facebook and Instagram’s parent company disclosed that it has “contributed more than $100 million to programs supporting our fact-checking efforts since 2016.”

Despite this investments, a Harvard study claimed that “influence operations" impacted the 2022 Philippine elections as the presidential campaign became a “contest between two parallel political realities.”

Around 80 percent of Filipinos prefer to get their news from online sources, according to the Reuters Institute’s 2024 Digital News Report.

Among social media platforms, 61 percent of Filipino adults surveyed still prefer to use Facebook as a news source, the study showed.

REDUCING OVER REMOVING DISINFORMATION

While Meta has been removing 99 percent of terrorist videos on its platforms, it only seeks to “reduce” other disinformation content on its platforms instead of taking down all these materials as “removing content is not always a solution to an issue,” Gummi said.

“Transparency is a much more effective way to address the issue than completely eliminating the content,” she said.

“If you remove it, nobody knows what it was. There is value in users noting this because they are not just using one of our platforms,” she said.

“People should know what is wrong in order for them to be informed.”

While Meta platforms “have dedicated teams” to review content flagged as possible misinformation or disinformation posts, it is still challenging for social media companies to address all issues as “what is true and what is false is subjective sometimes,” Gummi said.

“We are not the arbiters of truth and don't think any single actor should be. It would require us to know every single truth about every single thing in the world — which is impossible,” she said.

“Not everything is simply true or false… it’s really not easy for us to determine is something is purely true or purely false. There can be degrees of truth and opinions on what is the truth,” she said.

PARTNERING WITH REGULATORS

Meta has been partnering with election bodies, respected publishers and fact-checking institutions to label misleading or false content as fake news, Gummi said.

“We are in constant touch with regulators… to ensure that our products are aligned with the kind of regulation that they have committed,” she said.

“It’s an ongoing conversation and we also wanna make sure that we are aligned so we can address them together.”

In May, the Philippines’ Commission on Elections (Comelec) said it is looking to “prohibit the use of deepfake and artificial intelligence" in campaign materials for the 2025 midterm elections.

“The abuse of AI (Artificial Intelligence) technology and deepfake videos undermines the integrity of elections and the credibility of public officials, candidates and election management authorities,” Comelec chairperson George Garcia said in his letter to the agency’s Commission en Banc.

So far, the poll body has yet to implement the policy, and say how it intends to police violators.

Meta said for now, it has been placing labels on materials created through AI.

“We have labels saying ‘Generated by Meta AI.’ There is a disclosure,” Gummi said.

“All of these would inform the users on consuming that information and deciding whether they should share that information or not,” she said.

“This is an evolving space and we are working with a lot of different experts and partners.”

Meta said it has been flagging fake news on its platforms with the help of 90 fact-checking partners, covering 60 languages in 37 countries.

“Harmful ads will be removed always… We are working on the technology to be able to regulate videos as well,” Gummi said.

“We will tell you that this violated this policy so please don't share it again. We notify users when they are sharing false news,” she said.

“We have filters and managers, not just saying that this is rated false, but we also provide a link to a better article,” she said.

Meta is not keen on taking sides in political debates and squabbles on its platforms, Gummi said.

“It’s not about one side or another. They have to be policy compliant… We do it in a neutral, global fashion,” she said.

The “problem is false news and how people consume it,” the Meta official said.

“People do not really understand how to filter news,” she said.

“We acknowledge that that is a huge responsibility on us.”