Limits to Control Workshop

June 11-13, 2025 · Louisville, KY

An in-person research workshop convened by participants to examine the theoretical limits of AI control. Hosted by HΩ at their request.

Visit website ↗ Read the statement ↗

About the workshop

The Limits to Control Workshop convened researchers to map the boundaries of AI control, defined as the reliable capacity to constrain AI behavior to intended outcomes through technical and institutional measures. The program combined presentations, whiteboard discussions, and collaborative problem-solving to clarify viable research directions and impossibility results.

At a glance

FormatIn-person workshop

Presentations, whiteboard discussions, and collaborative problem solving.

FocusLimits of AI control

Foundational theory, control mechanisms, and boundary conditions for stability.

OutcomesCollective statement

A shared worldview and research agenda; papers are forthcoming.

Hosted by

Hosted at participants’ request; organized by Orpheus Lummis, Remmelt Ellen, and Thibaud Veron.

Proceedings

Paper publications are pending. For now:

Note: The statement reflects participants’ views.

Participants

Papers

Essays

Organizing team

Hosted by HΩ at participants’ request. Organized by:

Contact

For inquiries, contact Orpheus at o@horizonomega.org.