AI Transforms Israeli Elections, Raising Privacy and Authenticity Concerns

Israel's current election campaign marks the first time artificial intelligence (AI) is a significant tool for political campaigns, introducing complex questions about voter privacy and the authenticity of political messaging. AI's ability to rapidly generate realistic videos, images, and audio raises concerns about deepfakes and misinformation, while also enabling parties to process vast amounts of voter data to tailor messages to specific individuals or groups. This election cycle sits at the intersection of election law, privacy regulations, and rapid AI advancements.

Israeli law now requires clear labeling for propaganda content that uses "deepfakes" or digitally altered material designed to appear authentic. The goal is transparency, allowing voters to distinguish between real and synthetic content without outright banning AI in campaigns. However, labeling does not absolve parties from other legal obligations, such as privacy laws, especially when using individuals' likenesses, voices, or personal information.

Political parties can access voter registry information for campaign purposes, but this access is not unlimited. Direct mail based on personal characteristics triggers privacy laws, requiring disclosure of the information source and the option to be removed from mailing lists. The use of inferred data, where AI algorithms deduce preferences or behaviors from various data points, further complicates privacy issues, as the legality of using such inferred information depends on its source, purpose, and adherence to original collection intentions.

AI's capacity to precisely target messages based on inferred political profiles raises significant privacy concerns. While political segmentation is not new, AI amplifies the scale, accuracy, and automation of this process. Legal questions now focus on the data sources used for segmentation, the legality of their use, transparency, and whether usage exceeds the original purpose for which the data was collected.

Parties are also subject to data security obligations. Large databases, especially those containing information on a broad public, require robust security measures to mitigate risks and comply with legal requirements. A breach of such databases could lead to financial penalties and, in some cases, criminal liability. The increased number of involved parties during elections, including vendors and technology systems, heightens the need for strict access controls and oversight.

AI also presents a risk of foreign interference, as hostile actors can generate and disseminate large volumes of tailored disinformation rapidly. This is particularly concerning when combined with personal or group data, allowing for targeted influence campaigns. The speed of AI-generated content dissemination challenges the effectiveness of legal remedies, especially close to election day.

The article advises voters to develop healthy skepticism towards unusual content, especially close to elections, and to verify sources and authenticity. Voters are encouraged to question how parties obtain their personal information and to exercise their rights regarding direct mail. Ultimately, in the age of AI, personal data protection is intertwined with protecting a voter's ability to make independent decisions.