Significant action taken by Meta
Meta, the parent company of Facebook and Instagram, has announced its intention to assemble a dedicated team to combat deceptive artificial intelligence (AI) content ahead of the upcoming EU elections in June.
The company is particularly concerned about the potential misuse of generative AI, which can create fake videos, images, and audio to manipulate voters.
This announcement coincides with remarks from Home Secretary James Cleverly, who warned about the potential use of AI-generated fakes to influence general elections. However, some experts have expressed skepticism, suggesting that Meta’s plans may not be sufficiently robust.
The BBC has reached out to Meta for clarification on whether similar measures are planned for upcoming elections in the UK and US. Thus, we can expect the answer in the coming days,
Meta’s announcement follows its recent collaboration with other major tech companies to combat deceptive content.
The European Parliament elections are scheduled to take place from 6 to 9 June this year. In a parallel move, social media platform TikTok revealed plans to launch “Election Centers” within its app for each of the 27 EU member states, providing authoritative information in local languages.
Meta’s Elections Operation Centre
In a recent blog post, Marco Pancini, Meta’s head of EU affairs, announced plans to establish a dedicated “EU-specific Elections Operations Centre.”
This initiative aims to swiftly identify and address potential threats across Meta’s platforms, including Facebook, Instagram, WhatsApp, and Threads, during the upcoming EU elections.
Pancini emphasized Meta’s significant commitment to safety and security, amounting to more than $20 billion in investments since 2016. He also noted the significant expansion of Meta’s global team focused on these efforts, which now comprises approximately 40,000 individuals.
Among them are 15,000 content reviewers proficient in over 70 languages, including all 24 official EU languages. Pancini emphasized the collaboration of experts from various disciplines within Meta, such as engineering, data science, and law, to enhance the effectiveness of these measures.
Several challenges lie ahead
Deepak Padmanabhan from Queen’s University Belfast, co-author of a paper on elections and AI, pointed out shortcomings in Meta’s announcement, expressing concerns about the effectiveness of the planned strategy.
He highlighted issues with Meta’s approach to handling AI-generated images, which he deemed potentially unfeasible. Padmanabhan raised a scenario where realistic AI-generated images depict clashes between protesters and police, questioning the ability of technology or human experts to definitively label such content as fake or real.
He emphasized the uncertainty surrounding Meta’s generative AI strategy, citing serious limitations.
Despite Meta’s collaboration with fact-checking organizations across the EU, including the addition of three more partners, the focus remains on debunking misinformation rather than addressing content aimed at suppressing voting.
Meta’s measures involve labeling such posts, reducing their visibility, and prohibiting them from being used in ads. Ads are restricted from challenging the legitimacy of the vote, prematurely claiming victory, or questioning election methods and processes.
Comments 1