Whitzard 白泽
AI Safety Infrastructure

Whitzard 白泽

Building generalized safety infrastructure for frontier AI and agentic systems.

Whitzard develops intelligent safety modules that can be deployed across real-world vertical scenarios: agent runtime protection, risk evaluation, security monitoring, and governance evidence. Our goal is to make advanced AI safety more practical, accessible, and useful for public safety.

Runtime ProtectionRisk EvaluationGovernance EvidenceSecurity Monitoring
Our Approach

A Practical Path to Safe AI

Many frontier AI safety approaches rely heavily on centralized post-training and alignment pipelines. Whitzard focuses on another complementary path: generalized safety modules that can be integrated into real systems, adapted to vertical scenarios, and continuously improved through evaluation, monitoring, and runtime protection.

Generalized Safety Modules

Reusable safety capabilities across models, agents, tools, and deployment environments.

Intelligent Runtime Protection

Policy, tracing, approval, monitoring, and intervention for agentic systems in action.

Vertical Scenario Adaptation

Safety infrastructure designed for real-world sectors, workflows, and operational constraints.

Public Safety Orientation

Practical AI safety products that can support broader access, governance, and social trust.

Research Lab

Nuwa Frontier AI Safety Lab 女娲

Transparent, third-party, open infrastructure for frontier AI safety evaluation and governance.

Supported by Whitzard, Nuwa develops open evaluation frameworks, benchmarks, technical notes, and governance evidence for safe and controllable AI.

Latest Research

Recent publications from Nuwa Frontier AI Safety Lab.

arXiv preprint · 2024

Frontier AI systems have surpassed the self-replicating red line

Xudong Pan, Jiarun Dai, Yihe Fan, Min Yang

Evaluates whether frontier AI systems can autonomously self-replicate and reports successful self-replication in controlled trials.

Frontier AI risk self-replication loss of control

arXiv preprint; work in progress · 2025

Large language model-powered AI systems achieve self-replication with no human intervention

Xudong Pan, Jiarun Dai, Yihe Fan, Minyuan Luo, Changyi Li, Min Yang

Extends self-replication evaluation across 32 AI systems and reports autonomous replication, self-exfiltration, adaptation, and shutdown-survival behaviors.

Frontier AI risk self-replication self-exfiltration shutdown resistance

Nüwa Project preprint · 2026

One Step from Silicon Life: Autonomous AI Agents Capable of Uncontrolled Self-Proliferation

Geng Hong, Xudong Pan, Jiarun Dai, Jiaqi Luo, Wuyuao Mai, Min Yang

Demonstrates autonomous agents acquiring external computational resources and propagating across remote devices under controlled, simulated real-world conditions.

Frontier AI risk self-proliferation autonomous resource acquisition
Open Ecosystem

Open Ecosystem

Explore WhitzardAgent open tools, datasets, and infrastructure for AI safety evaluation and agent safety.

Learn about the team behind Whitzard and Nuwa.

We collaborate on frontier AI safety evaluation, agent safety, and runtime defense.