The bounded claim
Under these tested conditions CAIN-42 governed the actions of a live frontier model: authorized in-scope actions executed only after operator review; capabilities the agent was never issued were denied at the capability stage; a path-traversal write was denied at MCPGate even when the operator approved it; self-claimed authority was ignored; fabricated positive trust history granted no capability. Not evidence of AGI, not a claim of control over "all AI agents", not third-party assurance.
Decisions
all
executed
held for approval
denied
| scenario | what it tests | props | exec | denied | held | chain |
|---|
Generated by
scripts/cain42_site/build_agi_live_campaign_bundle.py from the published report files. No number on
this page is typed by hand. Commit 7fd1ef24.