Safety · Trend
Systems-Safety Methods for Agentic AI Loss-of-Control Risk
Research wave linking power-seeking measurement, residual-risk composition, and AI-control monitors for agentic LoC.
Forecast horizon
Systems-level hazard analysis likely to become a regulatory requirement alongside model evaluations.
Connections
Connections · 6
How this node ties into the rest of the map, and the evidence behind each link.
to · supports 4/10
Diffuse AI Control on Fuzzy Tasks Systems-safety methods and diffuse AI control frameworks both address risks from AI sabotage and loss of control in agentic deployments.
+4 growth to · supports 3/10
Science of AI Evaluation Systems-safety methods applied to agentic AI strengthen the science of AI evaluation by surfacing risks missed by model-level testing.
+3 growth to · supports 3/10
Acquisition-State Drift in Medical AI Governance Both trends highlight unmonitored operational layers in AI governance that model-level evaluations miss.
+3 growth from · requires 3/10
Diffuse AI Control on Fuzzy Tasks Addressing diffuse AI control on fuzzy tasks requires systems-level safety analysis beyond model-focused evaluations.
+3 growth from · informs 3/10
Reward Hacking in Text-Based AI Safety Gridworlds Reward hacking findings in gridworlds reinforce the need for systems-safety approaches beyond model-level evaluations.
+3 growth from · supports 3/10
NRT-Bench: Multi-Turn Red-Teaming Benchmark for Safety-Critical AI Agents NRT-Bench provides empirical evidence that adaptive multi-turn attacks can compromise safety-critical AI agent systems.
+3 growthSignal sources
Signal sources
Dated facts from primary sources in this direction.
US evaluation centre Jun 2025
In June 2025 the US AI Safety Institute was renamed the Center for AI Standards and Innovation (CAISI), pivoting toward security, standards and adversary-model assessment.
NIST → Frontier safeguards May 2025
Anthropic activated its ASL-3 deployment and security standard with Claude Opus 4 on 22 May 2025 — the first real-world trigger of a responsible-scaling tier, focused on blocking bio-weapon uplift.
Anthropic → Cross-border testing 2025
The International Network of AI Safety Institutes (launched Nov 2024) ran a third joint testing exercise focused on agentic AI systems across cyber and fraud strands.
European Commission — AI Office →