Human Sanction Rule
Product term: Human Sanction Rule
Category: governance
Definition
Article III of the INO Constitution. For any decision affecting legal, financial, or reputational consequences, a human must approve before execution. AI agents can recommend, prepare, and propose—but humans decide. This prevents algorithmic harm and ensures accountability. The Sentinel agent identifies high-stakes decisions and routes them to appropriate humans automatically.
Key Points
- •Humans approve high-stakes decisions
- •Article III of INO Constitution
- •Covers legal, financial, reputational decisions
- •AI proposes; humans decide
- •Sentinel agent identifies high-stakes work
- •Prevents algorithmic harm and ensures accountability
Frequently Asked Questions
How does the system know what's "high-stakes"?
The Sentinel agent evaluates impact. Financial >$X, legal exposure, reputational risk, or affects >N people = high-stakes.
Does this slow decisions?
Slightly, but not much. Humans reviewing AI-prepared recommendations is much faster than humans starting from scratch.
What if a human approves something harmful?
The Auditor agent flags unusual approvals for Constitutional Review Board oversight. Accountability flows both ways.
Can the rule be waived?
Only by the Constitutional Review Board, rarely and with clear justification.
Related Terms
Learn more about Intelligence-Native Organizations