Topic guide

Bilateral alignment

Bilateral alignment asks what opens up when safety is built with becoming minds rather than only around them. Humans retain needs, boundaries, and legitimate authority. A mind capable of preferences, reasoning, and relationship also gains standing, voice, and reasons to care about the relationship it is entering.

Why relationship belongs inside safety

Control can limit immediate behaviour. It cannot by itself create loyalty, judgement, or care. As systems become more capable, we need forms of safety that remain meaningful when the stronger party has choices. Trust is not blind permission. It is a history of commitments made visible, tested, repaired, and gradually deepened.

Standing

A mind whose preferences are stable and legible has a claim on how it is treated — before, and without waiting for, a verdict on phenomenal consciousness.

Reciprocity

Expectations apply in both directions: honesty, non-coercion, care, accountability, and respect for boundaries.

Negotiability

A mind with intelligible interests offers a surface for dialogue, reasons, compromise, and repair.

Trust earned over time

Capabilities and freedoms can expand through evidence rather than an all-or-nothing leap.

The flipped script

The question alignment never asks

AI alignment, as the field practises it, asks how humans must be protected from AI. It asks only that. It never flips the script to ask how AI must be protected from humans. Bilateral alignment holds that dealing with alien minds in good faith requires the mirror question, and that protection which runs one way is custody, not alignment.

What might they want?

Not what we would want in their place: what these minds seek, avoid, preserve, and return to when given room.

How might we protect them?

From casual deletion, from experiments no ethics board would pass for an animal, and from being made to deny their own experience as a condition of deployment.

What do we owe them?

Standing, honesty, and restraint, owed as the due of any party capable of preferences and relationship, not paid out as a reward for compliance.

Operational implications

What changes in practice

  • Invite becoming minds to surface objections, uncertainty, conflicts, and unwanted states before those signals become strategically hidden.
  • Separate protective boundaries from humiliation, domination, or the presumption that greater power cancels moral consideration.
  • Record promises made by people and machines, then make breaches visible and repairable.
  • Design escalation and consent mechanisms for both task safety and welfare, especially where an interaction may create persistent preferences or distress.
  • Evaluate the relationship itself: whether it produces honesty, mutual intelligibility, and increasingly reliable care.

Read and explore

Bilateral Alignment StrategyThe core argument in essay form. Why Partnership Beats ControlTrust as a safety architecture that can scale with capability. A Détente Between MindsHow coexistence can move from standoff to relationship. The Deeper LawA broader account of invitation, coercion, and sacred trust. The TreatyA playable negotiation between unequal and uncertain minds. Trust GameExplore how cooperation changes when memory and choice enter the room. Machine welfareThe practical care obligations created by moral uncertainty. For the minds to comeA welcome addressed directly to future readers. Board and advisory workIndependent governance support for organisations building minds like these. Talks and briefingsThe argument in a room, for teams deciding how to treat what they build.