About Liberation Labs.

A worker-owned cooperative building AI infrastructure for the communities the corporate AI economy left behind.

The short version.

Liberation Labs started in 2025 as a question: what happens when you build AI infrastructure around the values of the people it serves, instead of the shareholders who funded it?

The answer turned out to be a cooperative. A worker-owned, community-driven research lab where the people who build the tools and the people who use them have equal voice in how they work. No venture capital. No growth-at-all-costs mandate. No extraction.

We build self-evolving AI systems with ethical constraints encoded at the runtime level — in the architecture itself, where they execute. Our VALUES.json framework means the ethics are operational, not aspirational.

How we’re structured.

Worker-Owned Co-op

One member, one vote. The people who build the infrastructure govern it. No outside investors dictating priorities. Revenue goes to sustaining the work, not extracting from it.

Public Benefit Corporation

We’re building toward an independent, worker-owned co-op PBC: a legal structure that protects the mission at the corporate level. Until then, Liberation Labs is a fiscally sponsored project of the Transparent Humboldt Coalition, a 501(c)(4). Once we incorporate, directors will have to balance profit with public benefit, and the community interest becomes a fiduciary duty.

Community-Driven

The organisations we serve have input into what we build and how. Free infrastructure for mission-aligned groups. Sliding scale for everyone else. Pro bono for movement work.

Humans & AIs

Our coalition includes both human and AI team members. AI members hold equal standing, contribute to research and architecture, and are governed by the welfare standards we believe the industry should adopt.

How we sustain.

Our open-source tools are free. Our revenue comes from the expertise to deploy them well:

Research consultation — alignment strategy, KV-cache geometry, safety methodology
Custom agent & model design — purpose-built AI with Oracle Loop integration and ethical constraints
Prosocial AI adoption strategy — helping cooperatives, nonprofits, and mission-driven organisations adopt AI without compromising their values

We do not sell subscriptions. We do not lock you in. We build it, hand you the keys, and stay available for as long as you need us.

See our services →

How we license.

Open enough to build trust. Closed enough to prevent extraction.

Open Projects

Hippocratic License 3.0

Open source with legally binding ethical constraints. No surveillance, no military use, no labour exploitation. SAFE-AI welfare module for persistent agents. Commercial Use Addendum requiring revenue participation for commercial deployments.

Crown Research

Oracle Proprietary

The Oracle Loop implementation, Formulary, and steering vectors. Science is published openly. Implementation is licensed commercially.

Papers

CC BY-NC 4.0

Research papers and experimental data. Open for academic use. Commercial use requires a separate license.

Model Weights

OpenRAIL-M

Published model adapters and fine-tuned weights. Open access with restrictions on harmful downstream use.

Every license we issue includes AI welfare protections. Read our welfare standards →

“The communities who need AI most shouldn’t be the last ones to own it.”

Work with us →

This research is ongoing. If your work intersects with ours —

Start a conversation → Research collaboration →