# AI HANDOFF — CTS-OU Paper 09 v0.1

## 必須保留
1. MetaTheoryKnowledge != MoralImmunity。
2. Author / guardian / AI / ASI / existence-group all remain inside audit scope。
3. Ethical fluency != ethical reliability。
4. Self-critique != safety。
5. Responsibility acceptance != ex ante authority。
6. Low discernment != low subject standing。
7. Higher authority/irreversibility => higher discernment burden。
8. MDC is a minimum distinction set, not a worth test。
9. TRA requires asking/owning the terminal question; “no regret” is not the pass criterion。
10. NoRegret != MoralCorrectness; RegretLater != automatic moral failure。
11. Collective consensus != responsibility cancellation。
12. System-level responsibility and actor-level responsibility coexist。
13. The theory's authors are explicitly not exempt.
14. CTS-OU v0.1 is not claimed as final meta-ethics.

## Source integrity
- UTF-8 canonical source
- `$...$` / `$$...$$`
- no unicode_escape
- no renderer round-trip
- validate before checksum/package