Abstract
Two parties attempting to reach an agreement may each have private or confidential information that is key to reaching the agreement, but they may prefer to disclose such information to their opposite party only after mutual commitment. This disclosure describes techniques that enable negotiation and decision-making via interactions between large language models (LLMs) that act on behalf of the negotiating parties. Each party provides their confidential information to their LLM, which serves effectively as that party’s digital agent. The LLMs converse with each other with full context but only communicate a limited subset of information back to each party. The techniques enable the reaching of agreements while ensuring that the negotiating parties cannot access or later use information provided during negotiations. The techniques also enable a person to delegate costly decision making and social filtering to an agent that preemptively handles it on behalf of the person in a way that hews to preferences of the person.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Bires, Max, "Facilitating Agreements without Information Disclosure", Technical Disclosure Commons, (October 22, 2025)
https://www.tdcommons.org/dpubs_series/8758