Israel has mounted a systematic effort to influence how artificial intelligence chatbots respond to questions about the war in Gaza and the Israeli Defense Forces, according to a Calcalist report published by the Israeli financial and technology news outlet Calcalist. The initiative represents one of the more concrete known attempts by a government to directly shape the outputs of large-language-model platforms as those tools become primary information sources for millions of users worldwide.

The effort sits at an intersection that defense and security analysts have been tracking with growing concern: the use of generative AI as an information environment that state and non-state actors may seek to manipulate, much as social media platforms became contested terrain in the previous decade. For context on how digital footprints can inadvertently expose sensitive information, GDD has previously reported on fitness app Strava mapping the locations of American military personnel overseas — a reminder that information leakage and information shaping are two sides of the same technology-driven challenge.

Scope and Mechanics of the Program

According to the Calcalist report, Israeli officials and associated organizations have engaged in a coordinated effort to submit corrections, flag content, and interact with AI developers in ways intended to push chatbot responses toward narratives more favorable to Israel’s military conduct and strategic framing. The report does not specify which AI platforms were targeted by name in every instance, nor does it provide a precise breakdown of which government ministry or unit is leading the effort — officials have not confirmed the full organizational structure behind the campaign.

The Calcalist account describes the initiative as part of a broader Israeli information strategy that has adapted to the reality that younger audiences increasingly turn to AI assistants rather than traditional news sources when forming initial impressions of ongoing conflicts. The approach reportedly includes both direct outreach to AI companies and the systematic preparation of content designed to surface in the datasets or feedback loops those companies use to refine model behavior. The technical mechanisms by which large-language models absorb post-deployment feedback vary by platform, and the report does not claim any single method is uniformly effective.

Strategic Implications for the AI Information Environment

The disclosed effort underscores a fast-moving policy gap. AI developers have invested heavily in content moderation and model alignment, but the specific question of how governments — allied or adversarial — attempt to systematically shape model outputs through organized feedback or content campaigns remains largely unaddressed in public regulatory frameworks. If effective, such campaigns could influence how foundational facts about military operations, casualty figures, or legal characterizations are presented to users who may never cross-check a primary source.

The Israeli initiative is unlikely to be unique. Governments with strong information-operations capabilities — including several that compete directly with Western interests — have both the motive and the technical capacity to pursue analogous programs, whether or not those efforts have been publicly documented. That asymmetry gives states willing to engage AI platforms systematically a potential advantage in shaping baseline public understanding of contested events. The development reinforces arguments made by AI-governance researchers that training-data integrity and post-deployment feedback mechanisms need clearer third-party oversight. This dynamic also has direct relevance to defense planners: Microsoft’s strategic recalibration in China illustrates how geopolitical pressure is already reshaping where and how major AI infrastructure is deployed, complicating any assumption that AI platforms operate as neutral information utilities.

Calcalist did not report on any formal response from the AI companies involved, and it remains unclear whether the Israeli effort has produced measurable changes in chatbot outputs. Israeli government officials had not issued a public statement on the program at the time of the report’s publication.