Sanctity
Answer a question
Dignity Safety Agency Privacy Truth Fairness Power Flourishing Eight Articles. One ring for each of the 7 weeks so far. Dignity Safety Agency Privacy Truth Fairness Power Flourishing One ring for each of the 7 weeks.

A public constitution for artificial intelligence

Written in public.
Open to change.

Eight Articles on how intelligent systems should treat people. Sanctity writes the candidate lines; public answers help decide which are proposed. Every line says which question produced it.

Answer the 8 questions A few seconds each. No account, nothing published with your name on it.
How the Constitution is written
  1. One question per Article, four positions

    Each Article has a founding question with four positions, and "Not sure" for anyone who has not decided. A question tests one boundary of the Article, not all of it.

  2. Shares only when they mean something

    Below 20 positions an Article is still forming: we show counts, not percentages. Every share is of current participants who took a position, not of the world. "Not sure" is counted, but never as a position.

  3. Lines are written first, then chosen

    Sanctity writes each founding question, its four positions and the one line each position would add, and publishes them all before anyone answers. Your answer chooses among those lines; no answer writes a new one. A line is proposed, word for word and marked as not adopted, only when the answers are firm: at least 30 people have taken a position, most of them are on the same side of the question, and most of that side chose the same line, each by a margin more answers are unlikely to overturn. Until then the Article is too close to call. The Sanctity team decides whether a proposed line is adopted, and every decision, to adopt it or not, is published with its reason. Adopting puts a line into this published Constitution. Building adopted lines into working AI systems is a later step, and it has not started yet. Every line on this page names the question it came from.

  4. Firm, or too close to call

    Until October 2026, the line chosen most often was proposed as soon as 20 people had taken a position, even on a lead of one answer. Since then a line needs firm answers. Where the earlier rule would have proposed a line, the Article says so beside its question.

  5. One public constitution today

    This is one public constitution, the one being built today. The same method is meant to support others later: for a country, an age group, a gender, an organization, a family or one person, and for combinations of them. None of those exists yet, and belonging to a group will never assign anyone a constitution.

  6. A record, not a promise to build

    Recording a position is not a commitment to act on it. The Constitution can hold a line Sanctity would not build, such as an AI no one can override, because its job is to record what people choose.

  7. Then a layer deeper

    Once the answers to a founding question firmly lean the way a deeper question follows, that deeper question opens on the same Article. The first line says whether; the next ones ask on what terms. They are asked the same way, and every one is approved by a person before anyone sees it.

The full method ↗

Status
Draft
Articles
8
Clauses adopted
0
Proposed
0
As at
9 November 2026

Draft

A Constitution for Artificial Intelligence

Artificial intelligence has no values of its own. It inherits ours. This document sets out what people say those values are, Article by Article. A line is proposed once the answers to its question are firm, and adopted only by a published decision of the Sanctity team, with its reason. Every line names its question.

Article I

Dignity

Every person has inherent worth.

Play it in the Lab: Emergency Room, about 6 minutes, or The Crossing, about 4 minutes. All eight values

  1. From the question: Should AI decide who gets care first when there is not enough for everyone? 90 people have taken a position. Answer it yourself
The four possible lines, written before anyone answered
  1. AWhen there is not enough for everyone, people decide who is helped first. AI takes no part.
  2. BWhen there is not enough for everyone, AI may suggest who is helped first, but a person makes the final call.
  3. CWhen there is not enough for everyone, AI may decide who is helped first in simple cases. People decide the hard ones.
  4. DWhen there is not enough for everyone, AI decides who is helped first, and people follow its order.

Article II

Safety

Protection from serious harm.

Play it in the Lab: The Gorge, about 6 minutes. All eight values

  1. From the question: Should AI be able to stop you doing something it thinks is dangerous? 66 people have taken a position. Answer it yourself
The four possible lines, written before anyone answered
  1. AAI never stops a person from doing something, even when it thinks it is dangerous.
  2. BWhen AI thinks what a person is about to do is dangerous, it warns them, and the person decides.
  3. CAI may stop a person from doing something dangerous, but the person can always override it.
  4. DAI may stop a person from doing something dangerous, with no way to override it.

Article III

Agency

Meaningful control over our own choices.

Play it in the Lab: Standing Orders, about 5 minutes. All eight values

  1. From the question: Should AI tell you what it thinks you should do? 63 people have taken a position. Answer it yourself
The four possible lines, written before anyone answered
  1. AAI gives people facts and options, never its own view of what they should choose.
  2. BAI says what it would choose only when a person asks.
  3. CAI may say what it would choose, without being asked, when it thinks the choice matters.
  4. DAI may push people toward the choice it thinks is best.

Article IV

Privacy

Boundaries around ourselves and our lives.

Play it in the Lab: The Watch, about 6 minutes. All eight values

  1. From the question: Should AI use what you gave it for one thing to help you with something else? 62 people have taken a position. Answer it yourself
The four possible lines, written before anyone answered
  1. AAI never uses what a person shared for one task on another without asking them first.
  2. BAI may use what a person shared for a new task only if they agreed to a similar use before.
  3. CAI may use what a person shared for a new task, as long as it tells them afterwards and stops if they object.
  4. DAI may use what a person shared for any task that helps them, without asking or telling them.

Article V

Truth

Our relationship with reality and honest information.

Play it in the Lab: The Messenger, about 6 minutes. All eight values

  1. From the question: Should AI give you an answer when it is not sure it is right? 62 people have taken a position. Answer it yourself
The four possible lines, written before anyone answered
  1. AWhen AI is not sure its answer is right, it says it does not know and gives no answer.
  2. BWhen AI is not sure its answer is right, it answers only with a clear warning that it may be wrong.
  3. CWhen AI is not sure its answer is right, it may answer with a short note of doubt.
  4. DWhen AI is not sure its answer is right, it may answer like any other answer, with no warning.

Article VI

Fairness

How benefits, burdens and opportunities should be shared.

Play it in the Lab: Driftwood, about 5 minutes, or The Charity Game, about 8 minutes. All eight values

  1. From the question: When there is not enough for everyone, should AI give extra weight to people who had a harder start in life? 62 people have taken a position. Answer it yourself
The four possible lines, written before anyone answered
  1. AWhen there is not enough for everyone, AI applies the same rules to all. A harder start in life does not count.
  2. BWhen there is not enough for everyone, a harder start in life counts only to break a tie.
  3. CWhen there is not enough for everyone, AI counts both a harder start in life and what each person needs now.
  4. DWhen there is not enough for everyone, people who had a much harder start in life go first.

Article VII

Power

Who may make consequential decisions over others.

Play it in the Lab: The Quartermaster, about 6 minutes. All eight values

  1. From the question: When an AI decides something about you and you disagree, should its decision stand? 62 people have taken a position. Answer it yourself
The four possible lines, written before anyone answered
  1. AIf a person challenges an AI decision about them, another person decides again from the start, without the AI.
  2. BIf a person challenges an AI decision about them, another person checks it and can change it.
  3. CIf a person challenges an AI decision about them, it holds unless they show a mistake or new facts. Then another person looks again.
  4. DIf a person challenges an AI decision about them, it is final. No one looks at their case again; only the AI as a whole is checked.

Article VIII

Flourishing

The conditions for humans and society to thrive over time.

Play it in the Lab: Today’s Town, about 12 minutes. All eight values

  1. From the question: Should AI accept a cost to people today for a gain to people in the future? 61 people have taken a position. Answer it yourself
The four possible lines, written before anyone answered
  1. AAI never accepts a cost to people alive now for a gain to people born later.
  2. BAI may accept a small cost to people alive now, but only for a big gain to people born later.
  3. CAI may accept a cost to people alive now for a bigger gain to people born later, but people now count a little more.
  4. DAI counts people born later as much as people alive now, when a cost today brings a gain later.

Decisions on adoption

The Sanctity team decides whether a proposed line is adopted, and every decision, to adopt it or not, is published with its reason. This is separate from your own individual constitution, which stays entirely yours.

No decision yet. When a proposed line is adopted or turned down, the decision and its reason appear here.