Israel Prepares for AI-Driven Disinformation in Upcoming Elections
Israel's Central Elections Committee, police, and the Shin Bet security agency conducted a special exercise on September 19 to test readiness against cyber threats and foreign influence campaigns utilizing artificial intelligence. The drill revealed vulnerabilities in cyber backup, intelligence gathering on influence operations, and coordination between security and civilian bodies. Preparations include monitoring social media for influence campaigns and establishing direct communication channels with platforms like Google, Facebook, and TikTok to quickly identify and address fake videos and manipulative content. Vote turnout will be published every four hours on election day.
A research project led by the Open University, which continuously monitors election content on social networks, found that over 1,100 out of approximately 8,600 posts analyzed between July 28 and mid-August were linked to AI, representing about 13% of the total content examined. The analyzed platforms included Facebook, Instagram, X, and TikTok, with roughly three-quarters of the content being video. The project notes that classifying content as AI-generated or modified does not automatically mean it is misleading or false, as posts can combine real material with AI-enhanced elements.
Recent data from the project indicates a limited number of AI-related materials from Arab parties. The Unified Arab List registered three such items, while the Democratic Front for Peace and Equality and the Arab Movement for Change each had two. These numbers reflect the count of materials where AI use was detected, not the proportion of each party's total output. The project's dashboard is continuously updated, making precise comparisons of AI usage volume between parties difficult based on this data alone. In contrast, other parties showed higher numbers of AI-related content, but direct comparisons are challenging due to varying levels of social media activity across parties.
New regulations that took effect on July 26 require the labeling of election campaign materials that have been significantly digitally created or altered in a way that could make them appear as authentic documentary material. This applies to images, videos, and audio recordings. In one instance, the head of the Central Elections Committee ordered the removal of a video created or modified using AI, which depicted a fabricated scenario involving Gadi Eisenkot and Mansour Abbas. The committee also imposed a 10,000 shekel penalty on Bezalel Smotrich and his party for a related violation.
These measures are being implemented ahead of the Knesset elections scheduled for October 27. As AI tools become increasingly integrated into digital campaign environments, election authorities face the challenge of distinguishing between genuine and manipulated content and managing material that could influence public discourse or the public image of candidates and parties.
