A new team to be formed by Meta, the company behind Facebook and Instagram, to tackle the problem of fake AI content circulating during the European Union elections in June.
It is worried about how generative AI technology, which can create fake videos, images, and audio, might be used to deceive voters.
On the same day, Home Secretary James Cleverly said that some people will use fake AI technology to try to affect a general election.
However, a person who knows a lot about the industry said the plans might not be strong enough.
The BBC asked Meta if they have plans for the upcoming elections in the UK and US.
The announcement comes two weeks after Meta made a promise with other big tech companies to work together to stop harmful content.
The European Parliament will have a vote from June 6 to June 9 this year.
TikTok said it will add “Election Centres” in different languages within its app for each EU country to provide trustworthy information.
Marco Pancini, who oversees EU affairs for the company, said in a blog post that they will create a special center for dealing with elections in the EU. This center will watch out for possible problems and take action to fix them in their apps and technology right away. The company also owns WhatsApp and Threads.
“From 2016 until now, we have put more than $20 billion into making things safe and secure. We also have four times as many people, around 40,000, working on this. ”
This means there are 15,000 people who look at posts on Facebook, Instagram, and Threads in over 70 languages, including all 24 official EU languages.
He said this meant that experts from different teams in the company, like engineering, data science, and law, would work together.
‘Serious restrictions’
However, Deepak Padmanabhan from Queen’s University Belfast, who has written a paper on elections and AI, says that the announcement has some problems.
“He said that a lot of its planned strategy doesn’t have much substance. ”
He doesn’t like the way Meta plans to handle AI-generated images, and thinks it might not work.
He wanted to know what would happen if computer-made pictures showed protesters and police fighting.
“He said it’s difficult to prove the picture is fake because we need to make sure the attack didn’t happen. It’s hard for technology or experts to do. ”
“How can any technology decide if this is fake or real. ”
“So, we’re not sure how well Meta’s generative AI strategy will work – but there are definitely some problems. ”
Meta is working with 26 fact-checking organizations in the European Union. They are adding three more partners from Bulgaria, France, and Slovakia to help deal with the threat.
These organizations are not supposed to handle content that tries to stop people from voting. They are supposed to stop false information from spreading, even if it’s made using AI.
Mr Pancini said that these posts will have warning labels and won’t be shown as much. Also, they won’t be allowed in ads.
Ads cannot say the vote is not real, say they won before they did, or question how the election works.
He said the company’s work was a result of working together, and they would need to continue working together in the future.
He said that because AI-generated content is everywhere on the internet, we are also collaborating with other companies in our industry to create rules and standards.
This project is not just for one company. It will need a lot of work from different businesses, the government, and the public.