097
THE COMPANION
SECTION 18
ISSUE 001
PROJECTIONinteraction6-12mconfidence / high
Interfaces Graduate Agency
Projection: assistants will expose a visible ladder of agency—suggest, draft, simulate, act with confirmation, then bounded autonomy. The user should always see the current rung, permissions, and escape hatch. Agency is too consequential to remain an accidental side effect of connecting one more tool.
Why this idea is here
What the evidence establishes.
The agent index documents varying autonomy levels; hazardous-agent research demonstrates why action capability needs explicit safety boundaries.
Source ledger
Read the sources.
- S01The 2025 AI Agent Index — published June 2026
peer-reviewed transparency index / published 2026-06-25 / retrieved 2026-07-10
- S02AGENTSAFE: Benchmarking Embodied Agents on Hazardous Instructions
peer-reviewed benchmark / published 2026-06-01 / retrieved 2026-07-09