Meta Forms Team To Protect EU Elections From AI Risks

Facebook owner Meta said it plans to activate an operations centre to combat fraud and misinformation ahead of European Union parliamentary elections in June.

Competitor TikTok said earlier in February that it would similarly launch “Election Centres” in local EU languages within its app for each of the EU’s 27 EU members, to host authoritative information.

Meta said it has been preparing for the elections for some time and formed a dedicated team last year to develop a “tailored approach”.

Meta said it is currently expanding its fact-checking network with three new partners in Bulgaria, France and Slovakia.

Operations centre

Earlier this month the firm signed an accord along with many other tech companies vowing to work to make sure generative AI and other similar tools do not influence elections this year.

The EU-specific Elections Operations Centre is to bring together experts from across the company from disciplines ranging from intelligence, data science, engineering, research, operations, content policy and legal, Meta said.

Three areas of focus will be to combat misinformation, tackle organised influence operations and handle the influence of generative AI tools, Meta said.

The most serious type of misinformation is removed from Facebook as well as Instagram and Threads, which Meta also owns.

This includes content that could contribute to imminent violence or physical harm or is intended to suppress voting.

Fact-checking

For content that doesn’t violate those rules, the company works with 26 independent fact-checking organisation across the EU, covering 22 languages, who review and rate content.

If content is debunked by these fact-checkers, Meta attaches warning labels to it and reduces its distribution.

“When a fact-checked label is placed on a post, 95 percent of people don’t click through to view it,” said Marco Pancini, Meta’s head of EU affairs, in a blog post.

Meta labels photorealistic content created by Meta AI and is creating tools to label images from other major AI platforms.

It said it would also add a feature for people to disclose when they share AI-generated video or audio so a label can be added, and may apply penalties if people fail to do so.