Safety · Concept
Fairness as Symmetry Operation in ML
Formalizes ML bias as symmetry breaking and implements loss-based regularization as a symmetry-restoring mechanism, achieving over 90% violation reduction with ~5% accuracy cost.
Forecast horizon
Connections
Connections · 2
How this node ties into the rest of the map, and the evidence behind each link.
from · informs 4/10
StylisticBias: Visual Cue Social Bias Benchmark for MLLMs StylisticBias benchmark reveals concentrated visual cue biases in MLLMs, informing fairness measurement approaches.
+4 growth from · applies to 2/10
LLM Dialect Bias Against African American English (AAE) AAE dialect bias in LLMs represents a systematic fairness failure addressable through activation steering techniques.
+2 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 →