AI in the election campaign: 13% of politicians' content created with neural networks
During the current election campaign, parties and politicians have published over 1,100 pieces of content created using AI, accounting for about 13% of their total social media output. This is according to data from the "Zeh be-AI" (It's in AI) monitoring project led by the Open University.
In the current election campaign, parties and politicians have published more than 1,100 pieces of content created using AI, about 13% of all content they uploaded to social media platforms — this is according to data from a new tracking project led by the Open University, which identifies and centralizes all AI content published by parties and politicians. The leading party: Likud, whose politicians, against the backdrop of the primaries, are responsible for 70% of the AI content published.
The project, "Zeh be-AI" (It's in AI), was established to make the use of AI during the election campaign accessible to the public. It operates by scanning the Facebook, TikTok, Instagram, and X (formerly Twitter) accounts of politicians and parties, and analyzing the content they published — images, videos, posts, audio clips — in order to identify whether it is content created by AI. "We want to create public transparency tools to help the public deal with the entry of AI into the election system, and thus also help maintain the purity of the elections," the project leader, Prof. Anat Ben-David from the Open University, told Calcalist.
Verification Process
Verification is carried out through a multi-stage process. First, it is checked whether the account operator published a message clarifying that it is AI content (the Central Elections Committee required parties and politicians to attach such a message), whether the platform itself marked the content as AI, and whether there is a watermark of the system with which it was created. If one of these markings is not identified, the content is checked by AI recognition systems, whose decision is ratified by a human team. If the content is identified as an AI product, it is attached to the site's database along with relevant information such as the platform, type of content, and date of publication.
According to Ben-David, the uses range from simple posters to deepfake videos that could mislead voters. "The upcoming elections will be different from anything we have known until today, due to the central place that AI occupies in our lives," said Dana Mazia, manager of the Bright Initiatives project at Bright Data, which supports the initiative with technical tools.
Statistics and Trends
According to the project's data, as of yesterday morning, since July 28, parties and politicians have published 8,611 posts. 12.7% of them (1,097) were identified as having been created, at least in part, by AI. There is a trend of an increase in the use of AI content, both in absolute numbers and as a proportion of all content. The leading platform for publishing AI content is Facebook (32%), followed by Instagram (30%), X (20%), and TikTok (18%). 76% of the identified AI content were videos, and the rest were images.
Likud leads the AI content table with 773 pieces of content (70%), followed by a significant gap by the Democrats (5%), Yisrael Beiteinu (5%), Yashar (3%), Beyachad-Yesh Atid (3%), and Otzma Yehudit (3%). Ariel Kallner is responsible for 16% of AI content alone. The four leading politicians in the use of AI content are all from Likud: Ariel Kallner (179 publications), Hanoch Milwidsky (79), David Amsalem (66), and Avichay Buaron (61).
