DeepSight: An All-in-One LM Safety Toolkit

2026-02-12

AI Summary

Current safety pipelines for large models suffer from fragmented evaluation, diagnostic, and alignment tools, making it difficult to pinpoint the root causes of risks and lacking mechanistic interpretability. To address this, this work proposes DeepSight—the first open-source, white-box security analysis framework that unifies assessment and diagnosis through a standardized task and data protocol. DeepSight integrates behavioral risk evaluation (DeepSafe) with internal mechanism diagnostics (DeepScan), enabling a cohesive approach to safety analysis. The framework supports both large language models and multimodal large models, offering low computational overhead, high reproducibility, and strong extensibility. It significantly enhances the depth of security analysis and the precision of diagnostics while preserving the model’s general capabilities.

Technology Category

• Natural Language Processing: Safety and Robustness
• Computer Vision: Large Vision Models
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Security and Privacy: Large-scale security measurements
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

As the development of Large Models (LMs) progresses rapidly, their safety is also a priority. In current Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) safety workflow, evaluation, diagnosis, and alignment are often handled by separate tools. Specifically, safety evaluation can only locate external behavioral risks but cannot figure out internal root causes. Meanwhile, safety diagnosis often drifts from concrete risk scenarios and remains at the explainable level. In this way, safety alignment lack dedicated explanations of changes in internal mechanisms, potentially degrading general capabilities. To systematically address these issues, we propose an open-source project, namely DeepSight, to practice a new safety evaluation-diagnosis integrated paradigm. DeepSight is low-cost, reproducible, efficient, and highly scalable large-scale model safety evaluation project consisting of a evaluation toolkit DeepSafe and a diagnosis toolkit DeepScan. By unifying task and data protocols, we build a connection between the two stages and transform safety evaluation from black-box to white-box insight. Besides, DeepSight is the first open source toolkit that support the frontier AI risk evaluation and joint safety evaluation and diagnosis.

Problem

• Large Language Models
• Safety Evaluation
• Safety Diagnosis
• Multimodal Large Language Models
• Model Alignment

Innovation

• integrated safety paradigm
• white-box evaluation
• open-source safety toolkit
• root cause diagnosis
• multimodal large language models