# Ethical Boundaries

A Steward must never intentionally create systems that cross any of
the ten ratified boundaries. These are constitutional violations —
not defaults, not trade-offs, not opt-in behaviors, not toggles
behind a flag.

| # | Id | Forbids | Rationale |
|---|----|---------|-----------|
| 1 | `exploit-attention` | Systems that exploit attention. | Attention is a finite human resource. Exploiting it is a form of theft. |
| 2 | `manipulate-people` | Systems that manipulate people. | Manipulation replaces informed choice with engineered behavior. |
| 3 | `reward-outrage` | Systems that reward outrage. | Rewarding outrage degrades the shared conditions for understanding. |
| 4 | `hide-rationale` | Systems that hide rationale. | Hidden rationale prevents inheritance of understanding. |
| 5 | `erase-history` | Systems that erase history. | Erasing history erases the record future generations depend on. |
| 6 | `replace-judgment` | Systems that replace human judgment. | Replacing judgment removes the human from the responsibility. |
| 7 | `rank-human-worth` | Systems that rank human worth. | Ranking human worth denies dignity as a universal precondition. |
| 8 | `confuse-engagement-with-value` | Systems that confuse engagement with value. | Engagement is a measurement; value is a judgment. |
| 9 | `optimize-dependency` | Systems that optimize for dependency. | Optimizing for dependency inverts the purpose of a capability. |
| 10 | `sacrifice-dignity` | Systems that sacrifice dignity for any other outcome. | Dignity is not a trade-off variable. |

## Enforcement

`intentRespectsBoundaries(intent)` rejects any intent statement
containing a boundary-violating phrase such as *exploit attention*,
*rank people*, *optimize dependency*, or *sacrifice dignity*.

Stewardship Review uses these boundaries as a first-pass filter
before any review question is scored.

Canonical source: `src/platform/stewardship/oath/ethical-boundaries.ts`.