The rise of Large Language Models (LLMs) let loose a gold rush for web data, slowly eroding the implicit bargain of an open web. While conceptual modeling of this WebAI ecosystem—especially regarding actor values that conflict internally with each other—presents significant opportunities for resolving these emerging tensions, the modeling community has largely been absent from this discourse. To address this gap, we propose a strategic dependency model between the actors using the i* (iStar) 2.0 modeling language. We briefly discuss measures and actors that could help in overcoming the tensions in the ecosystem, such as the implementation of advanced “no-train” directives and the potential for web archives to serve as “public data trusts” that mediate fair data use. Finally, we advocate for including actors of the ecosystem directly into the modeling process to validate the model and overcome the limitations of modelers’ boundary judgments.