Formal verification framework for dataset lineage, policy compliance, and training-time safety guarantees.
EN: Safety and compatibility layer for non-DeepSeek models in DSH Web. 中文：面向 DSH Web 非 DeepSeek 模型的安全兼容适配层。
Safety-aware training and evaluation platform for sequential policies and frontier LLMs under a deterministic policy gate and strict verifier.
Abliteration of GPT Model - AI Governance Subject Study
Free AI contract review for models and talent — catches exclusivity traps, in-perpetuity image rights, upfront-fee scams, and commission stacking, then tells you exactly what to negotiate. Works with Claude, ChatGPT, Cursor, and 26+ tools.
Gray Swan — independent third-party profile of a public API surface, by API Evangelist. Gray Swan AI is an AI security company that helps enterprises deploy AI with confidence. Its Cygnal product is a real-time, drop-in secure proxy that fronts LLM providers (OpenAI, Anthropic, and Gemini request formats) with input/output filtering and threat moni
To associate your repository with the model-safety topic, visit your repo's landing page and select "manage topics."