Safety · Concept
In-Context Neurofeedback: Privileged Access Test for LLM Self-Control
Stricter neurofeedback setup shows LLMs fail to reliably control privileged internal representations, challenging prior metacognition claims.
Forecast horizon
Safety claims about model self-control will need privileged-access evaluation standards.
Connections
Connections · 3
How this node ties into the rest of the map, and the evidence behind each link.
to · informs 3/10
LLM Lie Detection and Deception Auditing Failure of privileged internal control weakens claims that LLMs have robust metacognitive access usable for safety.
+3 growth to · informs 3/10
Science of AI Evaluation Privileged-access redesign shows prior internal-control claims may reflect superficial prompt-inferable targets rather than metacognition.
+3 growth from · frames 3/10
The Veto Variable: Human Override as Goal-Independent Cost Both probe limits of assuming aligned goals or metacognitive self-control imply safe oversight behavior.
+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 →