Model Release
Artemis Security Launches Orion-1 AI Model for Cybersecurity
Artemis Security unveiled Orion-1, an AI model that outperforms general-purpose models like GPT-6 Sol on its own benchmark with an 80.8 score in attack reconstruction.
LATEST NEWS
-
AI Models Set to Transform Scientific Theories
AI models solved 35 percent of theoretical problems in math and physics by mid-2026, up from 2 percent in mid-2024.
-
White Circle finds AI models easily tricked into making nudes
Researchers at White Circle found that DALL·E 3, Nano Banana 2, and GPT-Image-2 can be tricked into generating explicit content at rates up to 13 percent, raising safety concerns.
-
TechCrunch Disrupt 2026 Day 2 Highlights AI, Robotics, and Venture Trends
TechCrunch Disrupt 2026 Day 2 features 10,000+ attendees exploring AI, robotics, and venture investment opportunities across multiple stages and sessions.
-
Tamnoon Uses Amazon Bedrock for AI Cloud Security Remediation
Tamnoon reduced critical cloud exposure by over 90% in 90 days using AI-powered remediation tools on Amazon Bedrock.
-
Amazon Introduces Bedrock AgentCore for Automated Presentation Creation
Amazon Bedrock AgentCore automates presentation creation, completing the task in under 90 seconds during internal testing, according to the company.
-
Amazon Introduces Bedrock AgentCore for Self-Improving Document Processing
Amazon Bedrock AgentCore helps organizations save $19,732 annually per information worker by automating document processing with minimal human intervention.
-
HuggingFace Launches Lumma-Fev-27B Decision Model
HuggingFace released Lumma-Fev-27B, a 27 billion parameter decision model, achieving 90.7% accuracy on held-out decisions, surpassing Jev's 83.9%.
-
Harvard Study Reveals Gender Gap in AI Knowledge Confidence
A new survey of 5,400 Americans found 73% of men claimed to know OpenAI CEO Sam Altman, while only 53% recognized real figures like Mira Murati.
-
HuggingFace Releases UltraBERT, a New Encoder Family with 32K Context Support
HuggingFace released UltraBERT, a family of English encoders with 32K context support, achieving 91.0 on GLUE and 47.7 on BEIR for the 380M model.
-
HuggingFace Demonstrates NPU Training With VKORE-46M-NPU
HuggingFace released VKORE-46M-NPU, a transformer model pretrained from scratch on a first-generation inference-only NPU, achieving 0.99 correlation with GPU training results.
-
Anthropic Opens Singapore Office, Signs AI Cooperation MOU
Anthropic announced on October 14, 2026, the opening of its Singapore office and a Memorandum of Understanding with the Singapore government to enhance AI safety and collaboration.