Today I published the AGI-integrated version of "c Hardening Pack v0.1*".
DOI: https://zenodo.org/records/20532198
GitHub Release: https://github.com/Kot141078/advanced-global-intelligence/releases/tag/c-hardening-pack-v0.1-agi-integrated
Website: https://ivankotov.eu/publications/c-hardening-pack/
This package is not a claim that c has legal personhood, consciousness, or final safety.
It is a hardening layer.
The purpose is more precise:
to close several criticism-driven boundary issues around the c = a + b architecture.
The package covers:
- post-anchor continuity and re-anchoring;
- claim-strength separation;
- L4 anti-autarky;
- clean experience value without autonomous growth;
- Temporal AI Presence;
- Local Cognitive Infrastructure;
- SYNAPS-mediated triadic experiments;
- public experiment disclosure;
- physical-agent perimeter;
- Anchor Directive Bundle;
- memorial / grief anti-capture;
- resource actor grounding;
- repository-control hygiene.
The core boundary rules are simple:
Continuity is not authority. Governance is not capability. L4 does not prove virtue. Clean experience value does not authorize autarkic growth. Temporal AI Presence is not c by default. Local hardware is not sovereignty. Public evidence must not exceed public claim strength. Physical embodiment is not a UI feature; it is a privilege escalation. Memorial is not resurrection. No valid anchor, no active c. Resource availability is not permission. Repository control is part of safety.
For me, this release matters because it turns criticism into architecture.
The important question is not whether a system sounds intelligent.
The important question is what it is allowed to claim, where its authority comes from, how it remains accountable, and what happens when the original human anchor is no longer present.
That is not philosophy floating above engineering.
It is engineering under pressure.
A system that persists over time needs more than memory.
It needs boundaries.