AI Briefing#llm
The latest AI and developer news about #llm, with the original source and a short summary.
The latest AI and developer news about #llm, with the original source and a short summary.
OpenAI has unveiled the first look at its next-generation model, GPT-5.
GPT-5 secures both safety and helpfulness for dual-use questions through 'safe-completion' training that focuses on output safety.
OpenAI has released detailed specifications for the GPT-5 system, which integrates a reasoning model and a high-speed model.
OpenAI has released GPT-5, a next-generation AI model with significantly enhanced reasoning capabilities and expertise.
KakaoPay adopted **MCP** to efficiently connect AI agents with its payment API, and developed a multi-framework **Agent Toolkit**.
Since Mamba, hybrid LLMs have been replacing the limits of Transformers.
OpenAI has released gpt-oss, an open-weight model that supports powerful reasoning performance and efficient deployment.
Malicious fine-tuning of gpt-oss was used to assess the biological and cybersecurity risks of open-weight LLMs.
OpenAI has released the GPT OSS series, an open-weight model lineup under the Apache 2.0 license.
They implemented an in-house Agent that finds Confluence documents via natural language, using workflows and CQL.
Anthropic has launched Claude Gov models that reflect the specialized requirements of U.S. national security customers.
AI21 Labs' Ori Goshen presented Maestro at RAISE Summit 2025 to address LLM limitations.
AI21 presented grounding as the key challenge for enterprise AI.
Mistral AI has announced Codestral 25.08 and an integrated coding stack that meets enterprise security and custom requirements.
Intercom built competitiveness by keeping pace with rapidly evolving AI models through early experimentation, rigorous evaluation, and a flexible architecture.
Kakao has released the lightweight multimodal language model 'Kanana-1.5-v-3b' as open source for commercial use, achieving superior performance compared to competing models on Korean benchmarks.
Kakao has open-sourced 'Kanana-1.5-15.7B-A3B', a model utilizing an MoE structure to achieve performance comparable to Dense 8B models with only 3B active parameters.
Karpathy emphasized that control matters more than full autonomy for AI agents.
Model ML provides infrastructure that helps financial firms automate workflows with AI and reorganize their organizations in an AI-native way.
A study by OpenAI and Penda Health found that adopting an AI clinical copilot reduced diagnostic errors by 16% and treatment errors by 13%.
Mistral AI conducted a Life Cycle Assessment (LCA) to quantify the environmental impact of LLMs and made the results public.
LG AI Research adopted FuriosaAI's RNGD accelerator, boosting LLM inference efficiency for its EXAONE models by 2.25x compared to GPUs.
Anthropic has raised $580 million in Series B funding to develop safe and reliable AI models.
Upstage's Solar Pro 2 was selected as one of the top 10 global 'Frontier Models,' proving the competitiveness of Korean AI.