x.com
Is Israel spending around $45 million to try and train AI chatbots in their narrative by usinng Clock Tower x?
Yes, reports from multiple outlets indicate that the Israeli government has funded a contract with Clock Tower X (run by former Trump campaign digital director Brad Parscale) valued at more than $45 million (commonly cited as ~$46.5 million total), and one explicit goal involves creating websites and content to influence how AI chatbots frame and respond on Israel-related topics. This is not a direct deal with OpenAI, Anthropic, Google, or other model providers to retrain their systems. Instead, it works indirectly by flooding the open web with optimized pro-Israel content (sometimes called “LLM poisoning” or GPT framing in the filings). The sites are designed more for search engines, crawlers (e.g., Common Crawl, which feeds many training datasets), and retrieval-augmented systems than for large human audiences. Key details from reporting Contract and parties: Israel engaged firms including Havas Media Network, which subcontracted to Clock Tower X (Parscale’s firm). It is registered under the U.S. Foreign Agents Registration Act (FARA). The work is described in filings as supporting a campaign to “combat antisemitism” and shape U.S. public opinion, particularly among younger/Gen Z and conservative audiences. Initial phases were around $1.5 million per month (totaling several million early on); it later expanded, with totals reported as exceeding $45 million (Wall Street Journal) or $46.5 million (Drop Site News, TIME, Quincy Institute/Responsible Statecraft, and others). Monthly rates have been reported as rising to ~$4.5 million in some accounts. AI-specific element: Contract language includes deploying “websites and content to deliver GPT framing results on GPT conversations” plus SEO efforts. Clock Tower X has stood up a network of sites (examples cited across reports include Allyvia.org, PaxPoint.org/Paxpoint, FactSignal.org, and others) that push narratives on topics such as the U.S.-Israel alliance, Israel’s conduct in Gaza, claims about journalists or specific incidents, peace efforts, and related issues. Content has appeared in Common Crawl crawls (hundreds of times in early-to-mid 2026 per some analyses) and has been cited in responses by systems like Perplexity, Microsoft Copilot, and others in testing. Parscale’s team has claimed success in getting AI systems to incorporate material from the sites. Broader campaign: The contract also covers AI-generated text messaging (e.g., via “Friends for Peace”-style outreach), conservative media partnerships, influencer outreach, social content (TikTok, Instagram, YouTube, etc.) aiming for tens of millions of impressions monthly, and other digital efforts. This sits within larger Israeli public diplomacy spending. Early 2025 reporting (e.g., Responsible Statecraft) highlighted an initial ~$6 million phase focused on Gen Z content plus the GPT framing language; later investigations (mid-2026, including WSJ, Drop Site, TIME) documented the scale-up, the website network, and observed effects on chatbot outputs. Snopes and similar fact-checks clarified that there is no evidence of a direct OpenAI/ChatGPT training contract—the influence is via publicly available web content. In short, the core of the claim holds: significant Israeli government funding (around the $45–46.5 million range) is going to Clock Tower X for influence work that explicitly includes efforts to shape AI chatbot narratives through web content. The effectiveness, exact ongoing totals, and full scope continue to be covered in investigative reporting based on FARA filings, contracts, and technical analysis of the sites’ visibility in training/retrieval data.
