Meta Decides To Form A Team To Prevent AI From Tampering With EU Elections

Facebook and Instagram owner Meta has decided to form a team to tackle the spread of misinformation through AI during the upcoming EU elections from June 6 to 9.

As the election approaches, we'll activate an Elections Operations Center to identify potential threats and put mitigations in place in real time.Marco Pancini, Meta's head of EU affairs

Meta already works with 26 fact-checking agencies across the EU, together covering 22 languages. As an additional step for the elections, the company has decided to add three more independent agencies from France, Slovakia, and Bulgaria.

The team will consist of experts from within the company, including engineering, operations, legal, content moderation, research, and data science professionals. Together, they will keep an eye on the type of content posted on their platforms and do their best to curb misinformation through AI abuse.

Read more: OpenAI takes stringent measures against election misinformation

More About Meta's Initiative

This announcement comes shortly after Home Secretary James Cleverly made a statement saying that a lot of people will definitely be using AI to influence the elections this year.

But this statement has nothing to do with Meta's initiative. This was planned a while ago and in fact, at the Munich Security Conference on 15th Feb, Meta along with 19 other big tech companies signed an accord, pledging to come up with better security steps that'll stop content edited with AI from tricking the voters.

This initiative was taken considering that this year, more than 4 billion people from across 40+ countries will be casting votes. Highly populated countries like India, the USA, and the UK are all having major elections.

The only shortcoming here is that they didn't say that the posts containing misinformation would be deleted. Instead, they are only planning to suppress its reach and put a label on it.

But on the brighter side, the company is also planning to set regulations for the ads posted on its platform during the election period. For instance, Pancini said that no ad would be allowed to prematurely guess the result of the poll or question the legitimacy of the process or method of the election.

Meta has been asked if they have similar plans for the elections in the USA and UK as well this year but the tech giant didn't make an official statement regarding this.

It's important to note that Meta isn't the only one taking steps to protect the elections from misinformation-OpenAI, Adobe, and Microsoft have also vowed to keep AI away from elections.TikTok also declared in February that each of the 27 EU members will get their own Election Centres" in the local language which they can use to share authentic information about the election.

Read more: FCC Bans AI-Generated Robocalls Ahead Of The US Elections

Meta Plans To Lable AI Images

Meta is also developing tools that'll automatically detect and label AI-generated pictures, especially the ones made using popular AI platforms like Adobe, Microsoft, Midjourney, and OpenAI.

The ultimate goal is to deploy these features across Facebook, Instagram, and Threads.

The company already does this for images generated through its own AI tools. Both invisible and visible markers are added to such content.

The only shortcoming right now is that there's no embedded indicator in AI videos and audio that Meta can use for labeling purposes. Also, seasoned hackers might try to remove the watermark for manipulated media content.

To counter this, Meta will soon release an option for users to disclose if they are using AI-generated content while posting and is working on making it harder for the miscreants online to remove the labels.

Why The Concern Over AI's Effect On The Elections?

The AI boom over the last couple of years has been a double-edged sword. While on one hand, it has benefited individuals and industries, on the flip side, it heavily fueled the spread of misinformation.

Now that you can create realistic pictures within seconds, just with a text prompt, it's much easier to influence the voters or tarnish the image of a candidate. For example, last year, a manipulated video of allegedly Joe Biden inappropriately touching his adult granddaughter went viral.

Content like this can negatively impact a candidate's chances at the poll. So, this initiative by Meta is a welcome step to keep the elections fair and help the voters make clear judgments, but it's unclear how effective it will be without an external verification system.

The post Meta Decides To Form A Team To Prevent AI From Tampering With EU Elections appeared first on The Tech Report.