ai-safety

Benchmarking Open-Ended Inference Optimization by AI Agents

Open-source Claude Code skills — 6 cognitive firewalls block AI hallucination, bias & sloppy reasoning. npx skills add

High-fidelity Claude Fable 5 (Mythos) environment emulation and automated multi-agent jailbreak (Pack Hunt) research laboratory.

一份来自2026年的AI精神病理学诊断报告 / A Pathological Diagnosis of AI Civilization

AI agents now operate with authority. Authority without discipline is how complex systems fail. Nuclear’s control loop, ported to AI-assisted software engineering.

Stop LLMs from hallucinating your guesses as facts. Clarity Gate is a verification protocol for your documents that are going to be provided to LLMs or RAG systems. Place automatically the missing uncertainty markers to avoid confident hallucinations. HITL for non-directly verifiable claims.

[CVPR2025] Official Repository for IMMUNE: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time Alignment

Deterministic AI defense framework: immutable input filtering, n-model cryptographic hash consensus, and self-learning adaptive rules. Zero dependencies. Hardware-sealed. Patent Pending.

Lightweight, Self-Hosted AI Guardrails Model based on ModernBERT.

Runtime detector for reward hacking and misalignment in LLM agents (89.7% F1 on 5,391 trajectories).

[ICML 2025] Vulnerability-Aware Alignment: Mitigating Uneven Forgetting in Harmful Fine-Tuning

Türkçe yapay zeka güvenliği için açık kaynak çatı: rehber serisi, uygulamalı akademi, araştırma & deneyler, 23+ araç ve 20 Hugging Face veri seti + model.

Single-agent multi-modal system for insurance damage claim verification. #1 globally, HackerRank Orchestrate June 2026 (15,295 registrants). Anthropic, OpenAI, Groq, and Gemini support with a deterministic safety gate

Tracking exactly what happens to the internal "circuitry" (induction heads) of a 2-layer attention-only Transformer when forced to undergo domain adaptation from prose to structured Python code.