AIOTOR principles · September 2026
AI should increase capability without obscuring responsibility.
AIOTOR designs intelligent features so that human and organizational authority remain explicit. AI assistance should not silently become the source of truth for consequential enterprise decisions.
Evidence before assertion
Where a product presents evidence-backed conclusions, the supporting context and uncertainty should remain reviewable. Generated explanations should be distinguishable from verified operational or scientific evidence.
Authorized purpose and scope
Intelligent systems and agents should operate within the purpose, data boundaries, permissions and controls approved for the deployment.
Scientific and medical use
Research and scientific products are intended to support investigation, education and evidence review. Medical or scientific conclusions requiring professional judgment should remain subject to qualified human review and the appropriate validation for the intended use.
Agentic systems
Autonomy does not imply authority. AI agents should have distinct identity, clear ownership, purpose-bound access, appropriate approval requirements and accountable outcomes. Protected data access should be denied by default unless it is explicitly authorized for an approved purpose. These principles apply to agent identity, authority, data access and accountable outcomes.