Ethos Kernel (research prototype)

Research kernel explored in simulation and internal tests (v0.0.0, Git install only — not PyPI). Open source (Apache-2.0); not a commercial safety or clinical product; DAO governance in-repo is mock.

Elevator summary

A behavioral prototype: multipolar ethical evaluation, Bayesian-style belief updates, narrative memory, and modules that humanize limits (weakness pole, forgiveness decay, persistence / “soul” backup metaphors) — implemented in Python with formal property tests.
An optional natural-language layer communicates and narrates without replacing kernel decisions; simulations and a browser dashboard make the model explorable without hardware.
Goal: transparent, inspectable research in computational ethics — useful for discourse, teaching, and staged collaboration — not a substitute for regulation, clinical judgment, or legal advice.

Due diligence & artifacts

Live site: https://mosexmacchinalab.com
Interactive demo: https://mosexmacchinalab.com/demo
Source & license: https://github.com/CuevazaArt/ethical-android-mvp (Apache-2.0)
Bibliography: https://github.com/CuevazaArt/ethical-android-mvp/blob/main/BIBLIOGRAPHY.md
Changelog / security: CHANGELOG.md · SECURITY.md

Contact (no public email)

Structured first contact reduces spam. Use GitHub Issues with the appropriate template:

Funding, partnership, or press
Technical or research question
Bug report
Security: see SECURITY.md (private advisories on GitHub when enabled).

Limitation

This is a research and educational prototype. It is not certified for safety-critical, clinical, legal, or compliance use. Do not rely on it as a substitute for human judgment, institutional review, or domain expertise.