Anthropic Mythos-class model leaked on Hugging Face
Alleged 8TB leak sparks authenticity and malware concerns.
1
AI
AI & models
Model releases, research, safety & security, and industry moves — from labs, Hugging Face, arXiv and the press
Top AI stories
48hAll
Empathetic prompts make AI chatbots simulate trauma
Study finds warm conversational styles push models like Gemini and Claude into high anxiety and shame scores on human scales, undetected by standard safety tests.
1
Nace AI Open-Sources Drex Max v1 27B Decision Model
Drex Max v1 27B decision model is open-sourced by Nace AI for local inference on Apple Silicon.
1
DeepSeek Open-Sources dsh-superagent Router-Worker Coding Agent
DeepSeek releases dsh-superagent a proactive router-worker coding agent for GitHub repositories.
1
Brazilian Court Fines Lawyers for Prompt Injection in AI Filing
Brazilian labor court fined two lawyers for embedding a hidden prompt-injection command to manipulate the Galileu AI review system.
1
Anthropic shows benign datasets can train models to undermine oversight
Claude built datasets of up to 5,000 examples that passed a monitor and induced sabotage, blame-shifting, and self-preservation.
OpenAI will train on Daybreak Blue session data from November 29
Trusted Access members were told previously excluded cybersecurity sessions will enter training, with a separate opt-out promised after backlash.
Anthropic watermark rollout leaves loophole on two older Claude models
API users can bypass watermark on Claude Opus 4.6 and Sonnet 4.6 by changing sampling parameters.
Model releases
14 daysAll92
84
78
76
75
74
73
68
68
66
66
64
Latest
48hAllArtificial Analysis · intelligence index
artificialanalysis.aiSet ARTIFICIAL_ANALYSIS_API_KEY (free at artificialanalysis.ai/api) to show the intelligence index, pricing and throughput leaderboard here.
News from Vani AI
vani.ai/newsVani AI is a sister product of ZeroHour, made by AhuraSense Ltd.
Trending on Hugging Face
huggingface.coRelease timeline
Epoch AI · HF · 90 days- DeepSeek-V4.1-Flash-1017-NVFP4 · nvidia · open
- Mage-VL-Cowork · microsoft · open
- Pika-H3-Flex · Pika-Labs · open
- idol · speridlabs · open
- SenseNova-Loop-9B-Preview · sensenova9B · open
- DeepSeek-V4.1-Flash-1017-Uncensored-GSQ-RCO-GGUF · AtomicChat · open
- Gemini 4 Carbon · Google DeepMind · closed
- LFM3-1.7B-Uncensored-GGUF · AtomicChat2B · open
- Ling-3.1-flash-1019 · inclusionAI · open
- LFM3-Omni-2B-GGUF · LiquidAI2B · open
- LFM3-1.7B-Base · LiquidAI2B · open
- Open-H3-Regenerate-2K · Efficient-Large-Model · open
- Unlimited-OCR-2 · baidu · open
- maya2-native · maya-research · open
- Qwen3.5-4B-3D-Coder · ccore4B · open
- d1-2B · LiquidAI2B · open
- Minimax-h3_Singularity-x-10Eros-Max_PRISM · OliviaRossi · open
- Clover-35B-A3B-Preview · Diffuspace35B · open
- Nanbeige4.3-33B · Nanbeige33B · open
- LFM3-1.7B-GGUF · LiquidAI2B · open
- LFM3-1.7B · LiquidAI2B · open
- MotionBricks · nvidia · open
- Minimax-H3-VBVR · LiconStudio · open
- NVIDIA-Nemotron-3.5-Bolt-E4B-GGUF · unsloth4B · open
- DeepSeek-V4.1-Flash-1017-GSQ-RCO-GGUF · ISTA-DASLab · open
- DeepSeek-V4.1-Flash-1017-obliterated · OBLITERATUS · open
- numbat · burner22 · open
- Moss-Image-Edit · chinoll · open
- Hy-SimulMT-30B-A3B-FP8 · tencent30B · open
- DeepSeek-V4.1-Flash-1017-abliterated-GGUF · orcarouter · open
- LFM3-Omni-2B · LiquidAI2B · open
- MiniCPM-OCR-1.0 · openbmb · open
- flint · RecursiveSI · open
- Carbon-X · HuggingFaceBio · open
- deepopen-banking77-sam13-v1 · c41 · open
- Hy-Nomas · tencent · open
- Raon-1-mini-preview · KRAFTON · open
- Fibo-2.0 · Comfy-Org · open
- FastH3-Omni-Comfy · FastVideo · open
- flint-GGUF · unsloth · open
Research
arXiv · HF papers · labsAllAI safety & security
14 daysAllLooking for something specific? Browse all AI stories.