Month: February 2024

NEW YORK — An oversight board is criticizing Facebook owner Meta’s policies regarding manipulated media as “incoherent” and insufficient to address the flood of online disinformation that already has begun to target elections across the globe this year. 

The quasi-independent board on Monday said its review of an altered video of President Joe Biden that spread on Facebook exposed gaps in the policy. The board said Meta should expand the policy to focus not only on videos generated with artificial intelligence, but on media regardless of how it was created. That includes fake audio recordings, which already have convincingly impersonated political candidates in the U.S. and elsewhere. 

It also said Meta should clarify the harm it is trying to prevent and should label images, videos and audio clips as manipulated instead of removing the posts altogether. 

The board’s feedback reflects the intense scrutiny that is facing many tech companies for their handling of election falsehoods in a year when voters in more than 50 countries will go to the polls. As both generative artificial intelligence deepfakes and lower-quality “cheap fakes” on social media threaten to mislead voters, the platforms are trying to catch up and respond to false posts while protecting users’ rights to free speech. 

“As it stands, the policy makes little sense,” oversight board co-chair Michael McConnell said of Meta’s policy in a statement on Monday. He said the company should close gaps in the policy while ensuring political speech is “unwaveringly protected.” 

Meta said it is reviewing the oversight board’s guidance and will respond publicly to the recommendations within 60 days. 

Spokesperson Corey Chambliss said while audio deepfakes aren’t mentioned in the company’s manipulated media policy, they are eligible to be fact-checked and will be labeled or down-ranked if fact-checkers rate them as false or altered. The company also takes action against any type of content if it violates Facebook’s Community Standards, he said. 

Facebook, which turned 20 this week, remains the most popular social media site for Americans to get their news, according to Pew. But other social media sites, among them Meta’s Instagram, WhatsApp and Threads, as well as X, YouTube and TikTok, also are potential hubs where deceptive media can spread and fool voters. 

Meta created its oversight board in 2020 to serve as a referee for content on its platforms. Its current recommendations come after it reviewed an altered clip of Biden and his adult granddaughter that was misleading but didn’t violate the company’s policies because it didn’t misrepresent anything he said. 

The original footage showed Biden placing an “I Voted” sticker high on his granddaughter’s chest, at her instruction, then kissing her on the cheek. The version that appeared on Facebook was altered to remove the important context, making it seem as if he touched her inappropriately. 

The board’s ruling on Monday upheld Meta’s 2023 decision to leave the seven-second clip up on Facebook, since it didn’t violate the company’s existing manipulated media policy. Meta’s current policy says it will remove videos created using artificial intelligence tools that misrepresent someone’s speech. 

“Since the video in this post was not altered using AI and it shows President Biden doing something he did not do (not something he didn’t say), it does not violate the existing policy,” the ruling read. 

The board advised the company to update the policy and label similar videos as manipulated in the future. It argued that to protect users’ rights to freedom of expression, Meta should label content as manipulated rather than removing it from the platform if it doesn’t violate any other policies. 

The board also noted that some forms of manipulated media are made for humor, parody or satire and should be protected. Instead of focusing on how a distorted image, video or audio clip was created, the company’s policy should focus on the harm manipulated posts can cause, such as disrupting the election process, the ruling said. 

Meta said on its website that it welcomes the Oversight Board’s ruling on the Biden post and will update the post after reviewing the board’s recommendations. 

Meta is required to heed the oversight board’s rulings on specific content decisions, though it’s under no obligation to follow the board’s broader recommendations. Still, the board has gotten the company to make some changes over the years, including making messages to users who violate its policies more specific to explain to them what they did wrong. 

Jen Golbeck, a professor in the University of Maryland’s College of Information Studies, said Meta is big enough to be a leader in labeling manipulated content, but follow-through is just as important as changing policy. 

“Will they implement those changes and then enforce them in the face of political pressure from the people who want to do bad things? That’s the real question,” she said. “If they do make those changes and don’t enforce them, it kind of further contributes to this destruction of trust that comes with misinformation.” 
…

Hong Kong — Scammers tricked a multinational firm out of some $26 million by impersonating senior executives using deepfake technology, Hong Kong police said Sunday, in one of the first cases of its kind in the city.

Law enforcement agencies are scrambling to keep up with generative artificial intelligence, which experts say holds potential for disinformation and misuse — such as deepfake images showing people mouthing things they never said.

A company employee in the Chinese finance hub received “video conference calls from someone posing as senior officers of the company requesting to transfer money to designated bank accounts,” police told AFP.

Police received a report of the incident on January 29, at which point some HK$200 million ($26 million) had already been lost via 15 transfers.

“Investigations are still ongoing and no arrest has been made so far,” police said, without disclosing the company’s name.

The victim was working in the finance department, and the scammers pretended to be the firm’s U.K.-based chief financial officer, according to Hong Kong media reports.

Acting Senior Superintendent Baron Chan said the video conference call involved multiple participants, but all except the victim were impersonated.

“Scammers found publicly available video and audio of the impersonation targets via YouTube, then used deepfake technology to emulate their voices… to lure the victim to follow their instructions,” Chan told reporters.

The deepfake videos were pre-recorded and did not involve dialogue or interaction with the victim, he added.
…