Yuvion LLM: An Adversarially-Aware Large Language Model for Content And AI Safety

2026-06-25

AI Summary

Current large language models exhibit insufficient robustness against strategic adversarial attacks involving planning, tool use, and multi-step reasoning. This work proposes the first end-to-end security framework that treats adversarial robustness and agent capabilities as co-primary optimization objectives, encompassing adversarially aware data construction, knowledge-enhanced pretraining, risk-aware fine-tuning, and safety-oriented reinforcement learning. Building upon this framework, we introduce YLRE, a comprehensive evaluation benchmark comprising 93 tasks, and release the Yuvion-8B model. Experimental results demonstrate that Yuvion-8B significantly outperforms larger-scale models such as GPT-5.4 and Qwen3-MAX across diverse safety-critical tasks, achieving state-of-the-art adversarial robustness while preserving strong general-purpose capabilities.

Abstract

As large language models are increasingly deployed in real-world systems, safety failures can still lead to harmful outputs and dangerous misuse. We argue that the essence of safety is adversarial: many failures arise not from natural inputs alone, but from strategic attempts to evade model policies and safeguards. However, existing general-purpose model development largely overlook this adversarial nature, and often remain insufficient for realistic safety scenarios involving planning, tool use, and multi-step reasoning, causing measured safety performance to overestimate real deployment robustness. To address this gap, we present Yuvion LLM, a large language model built for adversarially robust content safety and broader AI safety. Yuvion LLM treats adversarial robustness and agentic capability as first-class objectives. Its pipeline combines adversarially aware data construction, knowledge-enhanced continued pretraining, and policy-grounded multi-task safety post-training, including risk-aware supervised fine-tuning and reinforcement learning-based policy optimization, together with safety-aware agentic reinforcement learning for tool use and multi-step reasoning in complex safety scenarios. We further introduce the Yuvion LLM RiskEval (YLRE), a collection of 93 benchmarks across four evaluation categories, covering diverse open and internal evaluations with a focus on safety, adversarial robustness, and real-world capability requirements. Across these evaluations, Yuvion LLM demonstrates clear advantages on safety-focused benchmarks and particularly strong robustness under adversarial conditions, while maintaining solid overall capability. Notably, Yuvion-8B outperforms most state-of-the-art baselines, including substantially larger models such as GPT-5.4 and Qwen3-MAX, on several safety tasks.

Problem

• adversarial robustness
• AI safety
• content safety
• large language models
• safety failures

Innovation

• adversarial robustness
• AI safety
• agentic reinforcement learning
• safety-aware training
• risk-aware fine-tuning