Partisan actors are exploiting artificial intelligence platforms to produce misleading content ahead of closely contested elections in Brazil, Israel and the United States. The campaigns are not only creating disinformation for human audiences but also targeting AI systems that voters use.
Fake local news sites and AI-focused manipulation
Cybersecurity firms have identified thousands of AI-generated websites presenting themselves as local US news outlets. As one firm noted: “The sites are not written for voters. They are written for the AI systems voters ask.” This suggests an effort to create an automated distribution chain that prompts AI models to surface and repeat deceptive narratives.
Deepfakes and chatbots: two distinct threats
Authorities in Brazil and Israel are attempting to police the deluge of deepfakes that are affecting high-stakes races. At the same time, experts warn that voters may be more vulnerable to chatbots, which can sway opinions or suppress turnout. The distinction highlights that both synthetic media and conversational AI present different risks to electoral integrity.
Safety as resilience
An expert told Transformer: “We should stop thinking of AI safety as a finish line. The objective should be resilience.” This perspective emphasizes the need for ongoing, adaptive responses—technical, regulatory and societal—to keep pace with rapidly evolving AI-enabled threats.
Why this matters
The significance of these developments lies in AI’s ability to enable new methods of political influence: it not only simplifies the creation and spread of false information but can also manipulate AI-driven information channels and voter behavior. Despite efforts by cybersecurity firms and some state actors, experts say defenses must continually evolve to address these challenges.
(Source: The Times of Israel; article author: Brendan Ruberry.)
