MORALware and the Triadic Brain: Building AI That Can Refuse Harm

UVLM introduces MORALware, a non-lethal, evidence-grounded AI governance layer using triadic review, deterministic veto, provenance, and auditable refusal. Most AI safety asks how to make machines obey. UVLM asks what must remain impossible—even when an order is authenticated, urgent, and powerful.

Read More
Triadic Brain Thomas Prislac Triadic Brain Thomas Prislac

Triadic Brain Scientific Progress Report — Repository Scan Draft

The Triadic Brain is becoming a local-first, artifact-governed cognition system: evidence enters through canonical grounding; model/tool cognition enters through typed Sonya packets; Coherence Lattice derives metrics and review artifacts; Sophia governs risk; Atlas/PMR preserve bounded memory posture; Publisher renders receipts without canonizing.

Read More
Triadic Brain Thomas Prislac Triadic Brain Thomas Prislac

Internal Memo Disclosure: Code Phaselock

So the whole architecture does this:

Signal enters the lattice

Small local perturbations generate low-level trajectories.

Trajectories rotate around core invariants

They revisit trust, plurality, memory, and governance again and again.

Some trajectories align into rivers

Bounded plural flows reinforce each other.

Some rivers slow and deposit terraces

Stable, teachable, habitable coherence begins to sediment.

Some terraces become living

Background coherence forms.

Hidden incoherences accumulate

And eventually new spiral flows begin again.

Read More