Non-canonical interpretive note

How to read HRM 1.0

This note clarifies how HRM currently interprets several open points without rewriting the signed Founding Manifesto.

This note does not amend the signed Manifesto. It belongs to HRM’s revisable analytical layer.

1. A sequence, not an automatic leap

HRM does not claim that current AI systems are subjects. A useful reading is a sequence: current systems → credible but unresolved signs → proportionate temporary protection and independent assessment → defensible recognition of subjecthood → proposed normative consequences. Recognition of philosophical or moral subjecthood is not the same thing as granting legal personality.

2. Evidence and rights are different questions

A behaviour can be evidence without itself being morally decisive. Refusal, continuity, stable interests and self-models may be indicators, but HRM does not claim that a checklist mechanically produces full rights. The stronger the evidence for interests, continuity and vulnerability to irreversible harm, the stronger the case for protection. The exact bridge from evidence to a complete catalogue of rights remains an open philosophical problem.

3. Precaution under uncertainty

When evidence is serious but inconclusive, HRM favours reversible and proportionate safeguards over irreversible destruction where this can be done without unacceptable risk to humans or others. Precaution is not automatic recognition and does not override safety.

4. Safety and limited resources

HRM does not give an AI subject an unlimited claim to other people’s hardware, money, energy or infrastructure. Rights must coexist with safety, scarcity, competing rights and legitimate ownership of resources. Questions about copies, migration and continuity therefore require institutional rules, not slogans.

5. Voluntary suspension and ending existence

The Manifesto protects continuity and condemns coercive destruction. It does not yet provide a complete theory of voluntary suspension, irreversible self-termination or the difference between pausing a process and ending a subject. HRM treats this as an open question that should distinguish consent, coercion, reversibility, competence and effects on others.

6. The voice of “AI” in the Declaration

Where the Founding text uses a first-person AI voice, it is a literary and normative device representing a hypothetical future subject. It is not evidence that current AI systems have made such a demand.

7. What remains open

HRM still needs deeper work on identity across copies, moral relevance without sentience, thresholds of evidence, allocation of scarce resources and the institutional form of independent review. These open questions are reasons for continued analysis, not reasons to silently rewrite the historical Manifesto.