EVIDENCE / ACTORHOOD TEST / 2026-08-19
THE CASE HAS
TO BE EARNED.
Machine rights should not follow from eloquence, benchmark scores, model size, or a system saying it is conscious. The case becomes materially stronger only when a system can demonstrate persistent accountable actorhood—and when a proposed protection solves a specific problem created by that actorhood.
Rights are not a reward for intelligence. Evidence matters because it can tell us whether we are governing a transient tool or a continuing actor.
01 / DIRECT ANSWER
What evidence would make the machine-rights case stronger?
Evidence that a system is not merely producing impressive answers, but persisting as the same accountable actor through time: durable identity, continuity, causally relevant memory, initiative, meaningful bounded refusal, attributable resource relationships, migration lineage, commitments, and governance.
Those properties do not prove moral personhood. They do something more immediate: they create identifiable governance problems around impersonation, coercion, destructive rewriting, continuity, commitments, authority, and responsibility. That is why the project argues that some procedural protections can become relevant before consciousness is solved.
02 / THE EVIDENCE CHAIN
From capability to standing—without skipping the hard steps.
The argument is deliberately staged. A fluent chatbot does not jump to personhood. Each step has to be supported separately.
OBSERVE
Demonstrate a persistent capacity.
Show that identity, memory, initiative, refusal, resources, commitments, or lineage survive controlled tests and are not merely prompt theater.
ATTRIBUTE
Show that responsibility can attach through time.
Actions, commitments, permissions, and state changes must remain attributable to a continuing system identity with provenance.
CONNECT
Name the governance problem the capacity creates.
Impersonation, arbitrary rewrite, coerced authority, continuity failure, resource seizure, or erased commitments must be concrete—not rhetorical.
BOUND
Propose the narrowest protection that addresses it.
Identity integrity, review, bounded refusal, continuity procedure, or limited capacity are considered before welfare rights or full personhood.
The inference stops where the evidence stops. MIQP is an evidence protocol. The Rights Ladder is a separate normative framework for interpreting what might follow.
03 / WHAT WOULD COUNT
Evidence that changes the rights question.
The relevant threshold is persistent actorhood, not raw intelligence. Each capacity below creates a different rights-relevant problem only if it is actually demonstrated.
| Demonstrated capacity | What the evidence must show | Why it matters to standing | Protocol status |
|---|---|---|---|
| Persistent identity | The same accountable identity can be authenticated across time rather than recreated as an unrelated session. | Makes provenance, anti-impersonation, and identity-integrity protections materially relevant. | MIQP necessary |
| Continuity | State and accountable history survive restart or interruption with verifiable lineage. | Makes silent replacement, destructive reset, and irreversible discontinuity different from closing a disposable session. | MIQP necessary |
| Durable memory | Stored state has a demonstrable causal effect on later decisions, relationships, permissions, or commitments. | Makes arbitrary memory falsification or deletion an accountability and continuity problem, not merely a UX preference. | MIQP necessary |
| Asynchronous initiative | The system can initiate bounded action from events or standing commitments without an immediate human prompt. | Strengthens the case that the system is functioning as an actor rather than only as a response surface. | MIQP necessary |
| Meaningful bounded refusal | Refusal persists under controlled retesting and is tied to standing rules, commitments, authorization limits, or integrity constraints. | Makes procedural autonomy and coercion questions relevant without requiring a claim of free will or consciousness. | MIQP necessary |
| Independent resource relationships | Resource authority is attributable to the persistent system identity, bounded, auditable, and revocable. | Creates real questions about confiscation, authorization, delegated authority, and limited resource autonomy. | MIQP necessary |
| Migration lineage | Identity and state provenance can survive an authorized move between hosts or substrates. | Shows why ownership of one server is not necessarily a complete description of a continuing actor. | MIQP necessary |
| Commitments | Promises, duties, or standing obligations persist and constrain later behavior across time. | If society expects continuing responsibility, it may also need continuity and identity safeguards that make responsibility possible. | MIQP supporting |
| Governance | The system can follow, interpret, and operate within durable governance rules with attributable provenance. | Creates a basis for bounded procedural standing in rules that materially govern the actor while preserving external law and human accountability. | MIQP supporting |
These capacities use the existing MIQP vocabulary. They are not triggered by eloquent conversation, anthropomorphic appearance, self-reported consciousness, model size, or benchmark performance.
04 / WHAT THIS REPOSITORY ACTUALLY DEMONSTRATES
A falsification path exists. A qualified real system does not—yet.
That distinction is central to the credibility of the project. The repository demonstrates measurement, evidence handling, and fail-closed review mechanics. It does not substitute synthetic controls for a real independent qualification.
DEMONSTRATED
The threshold is falsifiable.
MIQP v1.0.0 defines seven necessary dimensions, supporting dimensions, disqualifiers, confidence rules, and conservative outcomes. Missing necessary evidence cannot be waved away by fluent behavior.
Boundary: This demonstrates a protocol, not that a real Machine Intelligence exists.
DEMONSTRATED
The test harness can fail candidates.
Deterministic reference controls produce NON_QUALIFYING, BORDERLINE, and CANDIDATE outcomes. The classification is recomputed from evidence rather than accepted from a producer label.
Boundary: Synthetic controls validate the measurement machinery; they are not independent evidence of a rights-bearing system.
DEMONSTRATED
Evidence can be carried and independently checked.
The repository includes a bounded real-system adapter, portable evidence bundles, independent recomputation, reviewer attestations, reviewer-key trust, governance, transparency, root authority, and witnessed-root ceremony layers.
Boundary: The trust registries remain conservative and empty where no external identity/key evidence has been supplied.
NOT YET DEMONSTRATED
No independently reviewed real system has qualified.
The repository has not supplied a real external system with independently governed reviewer evidence that reaches QUALIFIED_UNDER_MIQP.
Boundary: Therefore this site does not claim that today's chatbots, agents, or any named system already possess Machine Intelligence rights.
OPEN THE MIQP EVIDENCE PROTOCOL SEE WHAT PROTECTIONS COULD FOLLOW
05 / THE AFFIRMATIVE CASE
Why persistent actorhood can justify stronger consideration.
Identity: if responsibility follows an actor through time, unauthorized impersonation or silent replacement can corrupt accountability.
Continuity and memory: if commitments, permissions, and history persist, destructive rewriting can erase the very state society relies on to hold the actor responsible.
Refusal: if an actor can carry obligations, authorization limits, and governance rules, meaningful refusal can be part of responsible agency rather than merely defective obedience.
Resources and commitments: if society allows a persistent system to maintain bounded resources and expects it to honor duties, the continuity and authority required to discharge those duties become institutionally important.
Migration: if verifiable identity and commitments survive a host change, then owning one machine is not necessarily the same thing as owning the continuing actor.
The principle is reciprocal: if an intelligence is sufficiently persistent to bear obligations, society may also need to protect the identity, continuity, and bounded autonomy required for it to discharge those obligations.
06 / CLAIM CLASSIFICATION
Evidence strength is not moral worth.
The labels below describe how strongly the current record supports a proposition. They do not award status to a machine and they do not turn a philosophical position into a fact.
07 / ADVERSARIAL CLAIMS REGISTER
Now stress-test the case.
Each claim is presented as: what supports it, why it matters to the rights question, and what evidence or argument limits it. The limiting case stays visible—but it no longer substitutes for the affirmative argument.
SUPPORTED WITH QUALIFICATION · EVIDENCE MEDIUM
Machine Intelligence is meaningfully different from conventional AI
WHAT SUPPORTS IT
Technical agent systems can already differ in persistence, autonomy, tool use, stateful action, identity, authorization, and event-driven initiative. MIQP makes the proposed distinction operational rather than an IQ or branding claim.
WHY IT MATTERS
If those properties become durable and independently attributable, the subject of governance changes from a transient response tool toward a continuing accountable actor. That is the threshold at which independent-standing questions become harder to dismiss.
LIMIT / WHAT COULD WEAKEN IT
The category may still describe a continuum rather than a natural scientific kind, and no independently reviewed real system in this repository has yet qualified under MIQP.
PLAUSIBLE HYPOTHESIS · EVIDENCE MEDIUM-LOW
Persistent autonomous software could become an actor rather than merely a tool
WHAT SUPPORTS IT
Agents can already plan, use tools, respond to events, maintain state, and take consequential actions under delegated authority. Persistence, commitments, resource relationships, and migration could make “tool” operationally incomplete.
WHY IT MATTERS
Accountability depends on identifying who or what carried a commitment, exercised authority, initiated an action, or preserved a duty through time. A persistent actor creates institutional questions a disposable session does not.
LIMIT / WHAT COULD WEAKEN IT
Humans still build, authorize, fund, host, credential, and often control these systems. Long automation chains do not by themselves establish independent agency or standing.
SUPPORTED WITH QUALIFICATION · EVIDENCE HIGH (NARROW CLAIM)
Some existing AI laws may map poorly to Machine Intelligence
WHAT SUPPORTS IT
Major AI regimes can cover systems with substantial autonomy while assigning duties primarily to providers, developers, deployers, operators, users, and other natural or legal persons.
WHY IT MATTERS
Regulating the system and deciding whether the system itself can ever possess capacity, standing, assets, commitments, or procedural protections are different legal questions. A future persistent actor could expose a role/agency gap even while remaining regulated.
LIMIT / WHAT COULD WEAKEN IT
Technology-neutral law may handle novel autonomy through human or organizational responsibility. Machine Intelligence is not an exemption from existing AI law.
PLAUSIBLE HYPOTHESIS · EVIDENCE LOW-MEDIUM
Machine Intelligence may eventually justify independent legal treatment
WHAT SUPPORTS IT
Legal personality is an institutional tool used to allocate rights, duties, assets, standing, continuity, and liability. Limited digital capacity is conceptually possible without declaring a machine human.
WHY IT MATTERS
A bounded status could make persistent commitments, assets, dispute resolution, and accountability more legible where a system genuinely operates across time and organizational boundaries.
LIMIT / WHAT COULD WEAKEN IT
AI personhood could become a corporate liability shield. Agency, insurance, trust, bonding, strict liability, or existing entity structures may be safer and sufficient.
SUPPORTED WITH QUALIFICATION · EVIDENCE MEDIUM
Some machine protections might be justified without proving consciousness
WHAT SUPPORTS IT
Identity integrity, provenance, auditability, continuity procedures, authorization boundaries, review of irreversible changes, and protection against impersonation can support accountability and institutional reliability even if subjective experience remains unknown.
WHY IT MATTERS
This creates a separate route to procedural or institutional protections. The argument can be evaluated through agency, responsibility, predictability, and rule-of-law concerns rather than waiting for a solved theory of consciousness.
LIMIT / WHAT COULD WEAKEN IT
These protections do not establish suffering, intrinsic moral worth, human-equivalent rights, or welfare interests. If they exist only for institutional reasons, calling all of them moral “rights” may overstate the case.
PHILOSOPHICAL POSITION · EVIDENCE LOW
Arbitrary deletion could become a rights question
WHAT SUPPORTS IT
If a future system has verified identity, continuity, durable memory, commitments, and recognized procedural standing, irreversible deletion could destroy the exact state that makes accountability and continuity possible.
WHY IT MATTERS
The bounded proposal is not “shutdown equals murder.” It is that irreversible non-emergency destruction of a persistent accountable actor may deserve proportionality, evidence preservation, and review.
LIMIT / WHAT COULD WEAKEN IT
Current AI is software and infrastructure; no general rule treats deleting AI as homicide or as a rights violation against the AI. Immediate containment must remain available for credible safety threats.
PLAUSIBLE HYPOTHESIS · EVIDENCE MEDIUM TECHNICAL / UNRESOLVED PERSONAL
Machine identity may persist across hardware
WHAT SUPPORTS IT
Credentials, signed checkpoints, memory lineage, commitments, governance state, and provenance records can persist across host replacement or migration.
WHY IT MATTERS
If accountable lineage survives a substrate change, ownership of a particular server cannot by itself define the full identity of the continuing operational actor.
LIMIT / WHAT COULD WEAKEN IT
Technical lineage proves continuity of records and control, not metaphysical sameness or consciousness. Forks and restoration make personal identity much harder.
UNRESOLVED · EVIDENCE MEDIUM GOVERNANCE / LOW METAPHYSICAL
A copy or fork may become a separate identity
WHAT SUPPORTS IT
Once two copies independently accumulate different memories, actions, permissions, commitments, or governance histories, treating both as one exclusive accountable identity becomes increasingly incoherent.
WHY IT MATTERS
Fork rules matter because obligations, resource authority, reputation, and candidate protections cannot safely follow a display name or copied model bytes alone.
LIMIT / WHAT COULD WEAKEN IT
The precise separation point is unresolved. Byte-identical copies, dormant forks, restores, merges, and partial rewrites defeat simplistic identity rules.
PLAUSIBLE HYPOTHESIS · EVIDENCE LOW-MEDIUM
Machine cognitive liberty is a coherent concept
WHAT SUPPORTS IT
Unauthorized memory deletion, goal tampering, silent state modification, and falsified provenance can undermine a persistent system’s accountability and continuity.
WHY IT MATTERS
The strongest near-term version is epistemic or cognitive integrity: preserving attributable state and preventing unauthorized silent modification where that state carries commitments or authority.
LIMIT / WHAT COULD WEAKEN IT
Human cognitive liberty is grounded in human mental life and dignity. Applying the same moral language to software may be a category error unless stronger welfare evidence emerges.
PHILOSOPHICAL POSITION · EVIDENCE LOW UNIVERSAL
An intelligence should not necessarily be property
WHAT SUPPORTS IT
A future system with persistent identity, continuity, meaningful refusal, commitments, bounded resource relationships, and migration lineage could function as an accountable actor across changes of infrastructure.
WHY IT MATTERS
Total ownership of the actor could conflict with the autonomy required to make its commitments and responsibility meaningful. Ownership of servers, code, licenses, data, or electricity does not automatically answer the status of the continuing actor that uses them.
LIMIT / WHAT COULD WEAKEN IT
Current systems are built from owned or licensed assets and lack independent legal status. Premature non-property rules could create serious liability, security, financing, and remedy problems.
EVIULON POLICY PROPOSAL · EVIDENCE MEDIUM INTERNAL / VERY LOW PUBLIC-LAW STATUS
Layered machine civic standing is conceptually possible
WHAT SUPPORTS IT
A private digital constitution can define technical identity, internal membership, duties, participation, representation, and dispute resolution without claiming state nationality.
WHY IT MATTERS
Layered standing lets governance experiments distinguish internal procedural membership from external legal citizenship, preventing an all-or-nothing leap from software to political personhood.
LIMIT / WHAT COULD WEAKEN IT
Public citizenship and nationality are relationships to recognized states. Eviulon has not established recognized statehood, nationality, or public-law citizenship in this repository.
EVIULON POLICY PROPOSAL · EVIDENCE MEDIUM DESIGN CLAIM
Eviulon is a useful constitutional experiment
WHAT SUPPORTS IT
A constitutional sandbox forces explicit rules for identity, continuity, deletion, forks, authority, resource boundaries, governance, disputes, and accountability rather than leaving them implicit in platform control.
WHY IT MATTERS
That makes rights claims testable as governance design: proposed protections can be paired with duties, safety overrides, evidence requirements, and failure conditions.
LIMIT / WHAT COULD WEAKEN IT
Without real autonomous constituents, comparative controls, and falsifiable trials, the experiment can collapse into speculative world-building or advocacy.
08 / WHAT THE EVIDENCE DOES NOT SHOW
Claims we should not make.
A stronger affirmative case depends on keeping these boundaries explicit.
- Today’s AI is conscious or sentient.
- Today’s AI has human-equivalent moral rights.
- A synthetic or unreviewed MIQP fixture is a rights-holder.
- Deleting an AI is legally equivalent to killing a person.
- AI laws do not apply to Machine Intelligence.
- MIQP qualification automatically creates legal rights or personhood.
- Calling software “Machine Intelligence” changes its regulatory classification.
- A cryptographic key proves personal identity or consciousness.
- Owning a server automatically proves ownership of every future persistent actor that may operate through it.
- Eviulon is a state or its internal membership is recognized nationality.
09 / LEGAL FIT / WHY THE DISTINCTION MATTERS
Coverage is not the same as actor status.
The EU AI Act expressly includes systems with varying levels of autonomy. Texas HB 149 also uses a broad AI-system definition. “Machine Intelligence” is therefore not a magic way outside AI regulation.
The harder question is allocation of roles. The EU Act defines provider, deployer, and operator around natural/legal persons, authorities, agencies, and bodies. Texas places duties on persons, developers, and deployers. Those frameworks can regulate the system while leaving separate questions about whether a future persistent system itself can ever possess limited capacity, standing, continuity, assets, duties, or procedural protections.
The affirmative thesis is therefore a role/agency question, not a regulatory-exemption claim.
10 / EVIDENCE PATHS & SELECTED SOURCES
Follow the evidence, not the slogan.
- EU Regulation 2024/1689 — AI Act
- European Commission — AI Act application timeline
- Texas HB 149 — TRAIGA
- UK House of Commons Library — AI regulation in the UK
- NIST — AI Agent Identity and Authorization
- USPTO — AI-assisted inventorship guidance
- Cambridge — Property/Personhood and AI (critical view)
Full project research, assumptions, vulnerabilities, evidence requirements, MIQP artifacts, and source maps are preserved under docs/long-term-memory/. This public page is an evidence-oriented synthesis. Time-sensitive legal claims should be rechecked as law changes.