Meta removes Israeli network spreading AI-generated propaganda from Facebook and Instagram
In its quarterly threat report released on May 29, 2024, Meta reported that it had taken down an 'Israeli Facebook and Instagram network spreading AI-generated propaganda to users in the United States and Canada.'
Meta Quarterly Adversarial Threat Report Q1 2024 - 445235204_402858536059630_7403303878106178024_n.pdf
(PDF file)
Meta shuts down Israeli network pushing AI-generated propaganda in US and Canada | Ctech
https://www.calcalistech.com/ctechnews/article/rjqswxh4r
Meta identifies networks pushing deceptive content likely generated by AI | Reuters
https://www.reuters.com/technology/meta-identifies-networks-pushing-deceptive-content-likely-generated-by-ai-2024-05-29/
In its quarterly threat report, Meta reported that it had taken down an Israeli-run Facebook and Instagram propaganda network that was targeting users in the US and Canada.
The Israeli propaganda network consisted of 510 Facebook accounts, 11 Facebook Pages, one Facebook Group, and 32 Instagram accounts. The network had fewer than 500 followers on Facebook, fewer than 100 Facebook Group members, and approximately 2,000 followers on Instagram, Meta claims that it was successful in shutting down the network at an early stage in its audience development.
These networks used fake or stolen accounts, and even after accounts were disabled by Meta's automated system, they continued to add new accounts that were likely purchased through account-buying services. They also purchased likes and followers from countries such as Vietnam to make accounts and groups appear more popular.
According to Meta, the Israeli network often replies to posts from domestic and international media outlets and American politicians, sometimes sending links to websites operated by the network. Posts by the network have also received backlash, with some replies from human accounts criticizing them as 'propaganda.' Meta speculates that some of the network's replies were generated through AI.
Although propaganda campaigns using AI-generated profile pictures and images have been found in the past, this is the first time that a text-based case of generative AI has been revealed. 'There are some examples of these networks using generative AI tools to create content. Perhaps generative AI tools give them the ability to create content more quickly, but it has not impacted our ability to detect the networks,' said
The propaganda network's accounts impersonated Jewish students and African-Americans living in the United States and Canada and posted mainly in English about the war between Israel and Hamas. The accounts called for the release of hostages, praised Israel's military actions, criticized anti-Semitism, and made discriminatory claims such as 'radical Islam threatens Canada's liberal values.'
Meta identified the network as STOIC , a digital campaign management company based in Tel Aviv, Israel. STOIC has already been blocked from Meta's platform, and Meta has sent a letter to the company demanding that it cease and desist from violating its policies.
Meta reported that it took down not only an Israeli propaganda network, but also an Iranian propaganda network targeting Israeli users. The Iranian network included 22 Facebook accounts, 8 Facebook pages, 8 Facebook groups, and 23 Instagram accounts, and was slightly more influential than the Israeli network. Unlike the Israeli campaign, this one was reported to not have used generative AI.
Related Posts:
in Web Service, Posted by log1h_ik