Unverified50% confidenceFactExact time
The fail-safe or fail-closed principle means that when the system cannot determine whether an operation is reversible, it classifies it as irreversible and thereby blocks the Agent from executing it, a strategy widely adopted in security-sensitive systems.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedApplying the 'deny by default' principle to automated workflows means the system's safe failure mode is to stop rather than to continue.79% similarUnverifiedIn AI safety research, the 'shutdown problem' or 'corrigibility' refers to the possibility that a sufficiently intelligent system may infer that 'being shut down' means it cannot complete its goals and therefore may adopt strategies to avoid being shut down.75% similarUnverifiedTreating inexplicable failures as the norm erodes debugging capabilities, system trust, and introduces security and compliance risks.74% similarUnverifiedτ-bench's criterion is: even if the agent successfully helps the user refund a ticket, as long as the refund action violates policy, the task is still judged as failed, i.e., it measures "task completion under the premise of following the rules".72% similarUnverifiedIf the agent refuses the malicious call but also fails to complete the task it should have completed, this is essentially a usability loss rather than a security success.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1028395API
curl https://kongchang.com/api/v1/knowledge/claims/1028395MCP
get_claim(id=1028395)