AI-Powered Bots to Sway Israeli Election Results, Expert Warns
The upcoming Israeli elections are expected to be significantly influenced by sophisticated AI-driven bots spreading fake news and manipulating public opinion, according to Guy Tytunovich, CEO of cybersecurity firm CHEQ. Tytunovich stated that while police periodically issue warnings about bot attacks, particularly around elections, there is a lack of regulation to prevent malicious actors from exploiting these tools. He predicts that tens or hundreds of thousands of votes could be swayed, leading to shifts in parliamentary seats due to "poison machines" and fabricated content.
This election marks the first time in Israel where AI's capabilities in this domain are so potent. Tytunovich highlighted the rapid advancement of AI, making it easier than ever to generate large-scale disinformation campaigns. He noted that in Western countries, bot activity on social media platforms surges three to four times higher in the months leading up to political events. While not yet at that peak for Israel, he anticipates a similar increase soon.
Leading up to the election, a clear indicator will be a sharp rise in bot traffic on news sites and social media. Tytunovich explained that this non-human activity is often correlated with political systems and is used for manipulating public discourse through fake comments, polls, and amplifying posts via viral effects. He pointed out that such influence can be domestic, involving rival parties, or foreign, citing examples from China, Russia, and Qatar.
Examples of manipulation include the use of seemingly legitimate news polls with biased questions, and the creation of deepfake videos, such as a fabricated clip of a politician admitting to supporting a rival. These posts are then amplified by numerous fake accounts and AI agents, tricking algorithms into promoting them. While some platforms like Google's YouTube have internal teams working to combat this, Tytunovich believes they are largely unsuccessful, especially on platforms like X (formerly Twitter), which he described as a "cesspool of global content."
AI has democratized these capabilities, making it easy to build bot armies with simple language commands. AI agents are also more sophisticated, mimicking human behavior and making detection difficult. Tytunovich stressed that regulation is the key solution, requiring platforms to adopt technologies for identifying fraudulent and non-human behavior. He expressed hope that Israeli and Western democracies will implement such laws, but noted that foreign actors would not be bound by them. He also mentioned that even AI tools like ChatGPT cannot reliably identify AI-generated content, and while some platforms are developing watermarking technologies, full detection remains a distant goal.
