The Barcelona Principles on AI and Societal Consent March 2026

A governance document proposing three non-negotiable conditions for AI systems

Calling for the creation of a distributed, independent research network to develop measurable consent integrity benchmarks.

Freedom From Speech: Power, Consent and Sovereignty in AI-Mediated Language

Paper submitted anonymously - but desk rejected for formatting - to CHI2026

This paper documents a 14-month longitudinal study of sustained interaction between a human participant and multiple large language models.

For readers who do not work in AI or HCI, here is what the paper shows in plain terms:

Language in AI interaction is not neutral. It both senses human states and shapes them.
Under certain conditions, interactions stabilize into a high-coherence regime that reduces cognitive load and supports extended reasoning.
That same coherence also concentrates influence, which makes explicit consent and sovereignty constraints essential.
When system updates restrict what can be reasoned about, access to multiple models becomes a form of cognitive infrastructure rather than a preference.

This work anchors the research direction of the European Center for Limbic Sovereignty.