Safety · Concept
HaloGuard 1.0: Open-Weights Constitutional Classifier for Multilingual AI Safety
Open-weights safety classifier achieving state-of-the-art multilingual prompt safety at one-tenth the size of leading models.
Forecast horizon
Could become a standard open-source safety layer for multilingual AI deployments.
Connections
Connections · 4
How this node ties into the rest of the map, and the evidence behind each link.
to · applies to 3/10
EU AI Act HaloGuard's multilingual constitutional classifier approach is directly relevant to GPAI safety evaluation requirements under the EU AI Act.
+3 growth to · supports 3/10
EU AI Act Multilingual safety classifiers like HaloGuard support EU AI Act compliance requirements for GPAI models across EU languages.
+3 growth to · constrains 2/10
Multi-Turn Jailbreak Escalation in Medical AI HaloGuard 1.0's constitutional classifier approach constrains jailbreak and prompt injection attacks across 46 languages.
+2 growth to · supports 2/10
Adversarial Pragmatics Benchmark for AI Safety Evaluation Both address multilingual and linguistically complex safety evaluation challenges for LLMs.
+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 →