Artificial intelligence has found a new battleground: the American public’s social media feed. A detailed investigation by Calcalist Tech reveals that an Israeli government-linked operation has been deploying AI-generated content at scale to shift opinion among American audiences — a campaign that illustrates just how thoroughly generative AI is transforming the mechanics of modern influence operations.
The campaign, described in the report as backed by Israeli state-adjacent interests, uses AI tools to produce synthetic text, images, and social media posts crafted to appear organic. Rather than relying on human writers or traditional PR firms, the operation automates content production — effectively industrializing persuasion in a way that dramatically lowers costs while dramatically raising volume.
Automation at the Core of a New Persuasion Playbook
What makes this operation technically notable is its reported reliance on large language model-driven content pipelines — the same class of tools powering commercial chatbots — repurposed to generate culturally tuned messaging for U.S. audiences. The content is reportedly optimized to target specific demographics and platform algorithms, with messaging adapted across platforms including X (formerly Twitter) and Facebook.
The scale separates this from earlier, more manual covert campaigns. Where past influence operations depended on networks of human-operated fake accounts, AI enables a smaller team to generate and distribute far more content with far less overhead. That efficiency gap is exactly what makes AI-assisted persuasion so attractive — and so difficult to counter. Platform trust-and-safety teams built to detect human-run inauthentic behavior are now racing to adapt their detection models to AI-generated signals that mimic organic posting patterns with increasing fidelity.
What This Signals for Platform Integrity and Policy
The Calcalist Tech investigation arrives at a moment when governments worldwide are debating how to regulate AI-generated political content. In the United States, the Federal Election Commission has been slow to issue rules specifically governing AI-generated political advertising, and no comprehensive federal framework yet covers synthetic content in foreign-linked influence operations targeting domestic audiences.
The operational model documented here — state-adjacent funding, commercially available AI tooling, and demographically targeted distribution — is almost certainly not unique to this campaign. Security researchers have warned for over two years that the barrier to running a sophisticated influence operation has collapsed as generative AI matured. What once required a well-staffed disinformation bureau can now be approximated with API access and a modest budget. This campaign, if confirmed at the scale Calcalist describes, may represent one of the clearest documented examples of that shift moving from theoretical risk to operational reality.
The implications ripple well beyond any single geopolitical conflict. Platforms, policymakers, and civil society organizations are going to need detection infrastructure that matches the speed of generation — and right now, that gap is wide. As DeepSeek’s new model and other frontier releases continue to push down the cost and raise the capability of open-weight language models, the tools available to influence operators will only improve. The harder question is whether the defenses will keep pace — and whether the regulatory will exists to mandate them before the next election cycle makes the answer obvious.
For a broader look at how AI is reshaping sensitive institutional decisions, Future Wire’s earlier coverage of politicized research offices offers relevant context on the intersection of technology, government, and public trust.
