Video summary

El contrato para que la IA hable bien de Israel

Main summary

Key takeaways

News and Commentary

Overview

The video argues that Israel paid to influence how ChatGPT and other AI systems talk about Israel—not by “hacking” the AI models, but by funding and deploying third-party web content designed to shape AI outputs.

Key Points Raised

  • Public DOJ contract / campaign framing

    • The speaker claims there is a public contract filed with the U.S. Department of Justice describing “deployment of websites and content” to deliver “GPT framing results” within AI conversations.

  • Named contributor / company and amounts

    • The video alleges the contract was signed by Brad Parkscal, linked to Donald Trump’s 2020 campaign.
    • It further alleges Parkscal’s company is Clock Tower X, operated via a global agency called AAS Media in Germany.
    • The video points to a form indicating supervision by the State of Israel.

  • Purpose stated as combating antisemitism

    • The video claims the stated purpose in the paperwork is to combat antisemitism, while framing the overall effect as steering AI narratives about Israel.

  • Scale and method of content creation

    • The speaker alleges the operation built dozens of sites/pages from scratch (described as ~20 sites with many optimized pages per month and large numbers of content variants).
    • It implies AI systems later ingest or retrieve this content.
    • The approach is described as targeting search-engine rankings.

  • How the video says AI gets “recent” information

    • The video argues AI systems rely on internet sources for up-to-date answers rather than “thinking independently,” so they can reflect whatever content has been published and optimized.

  • Examples from sister campaigns and content style

    • The video cites Politico coverage of similar material.
    • It claims the articles are written in a way that aligns with what users/courts might prompt AI to produce—i.e., answers positioned around questions.

  • Evidence of distribution / detection

    • The speaker claims traces were found on platforms such as Google Gemini and Microsoft Copilot.
    • It also alleges Perplexity cited one of the sites as a primary source without disclosing the payer.

  • No training, but citation / search use

    • The video says it found no evidence that ChatGPT was trained on the material.
    • It argues that the content can still be cited when AI systems search for related information.

  • Financial and valuation claims

    • The video reports figures such as $15M received and $13M disbursed.
    • It cites a Wall Street Journal estimate of the entire package as $46.5M per year.

  • Disclosure issues

    • The speaker claims required legal disclosures of sponsorship appear on website pages but are not visible on social media and may not show up in AI responses.
    • This is presented as making it harder for users to know the origin.

  • Response to confrontation

    • The video says Parkscal did not deny the arrangement.
    • It cites an outlet such as the Washington Examiner, alleging Parkscal said everything was declared and nothing had been “discovered.”

Overall Claim

The video’s overall argument is that Israel-linked funding and large-scale content deployment were used to influence AI-facing narratives, making AI systems more likely to present Israel favorably. The emphasis is on shaping “framing results” rather than changing the AI model internally.

Presenters / Contributors

  • Brad Parkscal (named in the video as a signatory/figure tied to the alleged campaign)

Original video