https://axbrief.com/news/the-llm-eval-harness-exposing-confident-hallucinations-in-enterprise-ai-wmevii AX BRIEF ko 2026-08-15T21:34:48.125Z LLM 평가 하네스, 모델이 가장 확신할 때 틀린다는 사실 발견했다 https://axbrief.com/en/news/the-llm-eval-harness-exposing-confident-hallucinations-in-enterprise-ai-wmevii AX BRIEF en 2026-08-15T21:34:48.125Z The LLM Eval Harness Exposing Confident Hallucinations in Enterprise AI https://axbrief.com/news/claude-adopts-synthid-text-to-meet-eu-ai-act-transparency-standards-lm9wek AX BRIEF ko 2026-08-15T19:23:05.350Z Anthropic, Claude에 SynthID-Text 기반 워터마크 도입 https://axbrief.com/en/news/claude-adopts-synthid-text-to-meet-eu-ai-act-transparency-standards-lm9wek AX BRIEF en 2026-08-15T19:23:05.350Z Claude Adopts SynthID-Text to Meet EU AI Act Transparency Standards https://axbrief.com/news/glm-5-3-hits-84-5-on-cybergym-to-challenge-gpt-5-6-sol-481erl AX BRIEF ko 2026-08-14T23:19:37.827Z Z.ai, GLM-5.3 공개... 포스트 트레이닝 확장으로 사이버 보안 및 코딩 성능 강화 https://axbrief.com/en/news/glm-5-3-hits-84-5-on-cybergym-to-challenge-gpt-5-6-sol-481erl AX BRIEF en 2026-08-14T23:19:37.827Z GLM-5.3 Hits 84.5% on CyberGym to Challenge GPT-5.6 Sol https://axbrief.com/news/the-3-000-tps-benchmark-driving-kogs-gpu-optimization-lm9yfr AX BRIEF ko 2026-08-14T21:00:24.504Z 표준 GPU에서 3,000 TPS 구현한 Kog, LLM 30배 가속화 도전 https://axbrief.com/en/news/the-3-000-tps-benchmark-driving-kogs-gpu-optimization-lm9yfr AX BRIEF en 2026-08-14T21:00:24.504Z The 3,000 TPS Benchmark Driving Kog's GPU Optimization https://axbrief.com/news/the-10-gas-price-risk-facing-meta-google-and-amazon-ai-hubs-lmaev8 AX BRIEF ko 2026-08-14T18:14:40.605Z 아마존·구글·메타·MS, AI 전력용 천연가스 발전소 건설... 가격 3배 상승 가능성 제기 https://axbrief.com/en/news/the-10-gas-price-risk-facing-meta-google-and-amazon-ai-hubs-lmaev8 AX BRIEF en 2026-08-14T18:14:40.605Z The $10 Gas Price Risk Facing Meta, Google, and Amazon AI Hubs https://axbrief.com/news/microsoft-copilot-unifies-consumer-and-business-apps-to-fight-chatgpt-lmagci AX BRIEF ko 2026-08-14T15:38:49.077Z Copilot 앱 통합하는 마이크로소프트, 8월 18일까지 저성과 기능 삭제 https://axbrief.com/en/news/microsoft-copilot-unifies-consumer-and-business-apps-to-fight-chatgpt-lmagci AX BRIEF en 2026-08-14T15:38:49.077Z Microsoft Copilot Unifies Consumer and Business Apps to Fight ChatGPT https://axbrief.com/article/49317760 AX BRIEF ko 2026-08-16T10:53:31.847Z 초등 데이터만 학습한 LittleLearner, 사전 학습 필터가 능력의 천장을 결정한다 https://axbrief.com/en/article/49317760 AX BRIEF en 2026-08-16T10:53:31.847Z LittleLearner Proves Pre-training Sets the AI Capability Ceiling https://axbrief.com/article/graph-engineering-the-new-orchestration-layer-for-probabilistic-ai-agents-32544 AX BRIEF ko 2026-08-16T04:11:20.912Z 에이전트 설계, 단일 루프에서 그래프 기반 오케스트레이션으로 확장 https://axbrief.com/en/article/graph-engineering-the-new-orchestration-layer-for-probabilistic-ai-agents-32544 AX BRIEF en 2026-08-16T04:11:20.912Z Graph Engineering: The New Orchestration Layer for Probabilistic AI Agents https://axbrief.com/column/glm-5-3-gbd-5-6-saul-and-veo-3-1-fast-debut-260816 AX BRIEF ko 2026-08-16T00:06:45.719Z AI 추론 속도 경쟁 점화 — GLM 5.3부터 비오 3.1까지 신형 모델 대거 등장 https://axbrief.com/en/column/glm-5-3-gbd-5-6-saul-and-veo-3-1-fast-debut-260816 AX BRIEF en 2026-08-16T00:06:45.719Z GLM 5.3, GBD 5.6 Saul, and Veo 3.1 Fast Debut https://axbrief.com/article/why-cloudflare-is-pivoting-its-massive-infrastructure-toward-ai-agents-49310719 AX BRIEF ko 2026-08-15T21:04:07.889Z Cloudflare, 인프라 전문성 버리고 'AI 기능 나열'에 빠진 이유 https://axbrief.com/en/article/why-cloudflare-is-pivoting-its-massive-infrastructure-toward-ai-agents-49310719 AX BRIEF en 2026-08-15T21:04:07.889Z Why Cloudflare is Pivoting Its Massive Infrastructure Toward AI Agents https://axbrief.com/article/ai-math-performance-the-triumph-of-working-memory-over-reasoning-49312845 AX BRIEF ko 2026-08-15T20:50:43.311Z AI의 수학적 성과, 추론 지능보다 '작업 기억' 확장 결과 https://axbrief.com/en/article/ai-math-performance-the-triumph-of-working-memory-over-reasoning-49312845 AX BRIEF en 2026-08-15T20:50:43.311Z AI Math Performance: The Triumph of Working Memory Over Reasoning https://axbrief.com/article/why-ai-collaboration-is-shifting-from-commands-to-intent-centric-49309451 AX BRIEF ko 2026-08-15T18:53:21.191Z AI 협업 패러다임 전환: 명령어 기반 코딩에서 의도 중심 리더십으로 https://axbrief.com/en/article/why-ai-collaboration-is-shifting-from-commands-to-intent-centric-49309451 AX BRIEF en 2026-08-15T18:53:21.191Z Why AI Collaboration is Shifting from Commands to Intent-Centric Leadership https://axbrief.com/article/ballast-for-claude-code-the-plugin-that-systematizes-ai-workflows-32529 AX BRIEF ko 2026-08-15T14:08:32.048Z Claude Code 플러그인 'ballast' 공개, 작업 절차 시스템화 지원 https://axbrief.com/en/article/ballast-for-claude-code-the-plugin-that-systematizes-ai-workflows-32529 AX BRIEF en 2026-08-15T14:08:32.048Z Ballast for Claude Code: The Plugin That Systematizes AI Workflows https://axbrief.com/article/suno-studio-2-0-shifts-from-ai-generator-to-full-scale-daw-32520 AX BRIEF ko 2026-08-15T02:38:00.754Z MIDI 편집과 AI 채팅 결합한 Suno Studio 2.0, 단순 생성기에서 DAW로 확장 https://axbrief.com/en/article/suno-studio-2-0-shifts-from-ai-generator-to-full-scale-daw-32520 AX BRIEF en 2026-08-15T02:38:00.754Z Suno Studio 2.0 Shifts From AI Generator to Full-Scale DAW https://axbrief.com/article/claude-code-introduces-loop-engineering-via-goal-and-loop-32519 AX BRIEF ko 2026-08-15T02:23:57.721Z Claude Code, /goal과 /loop 기반의 루프 엔지니어링 공개 https://axbrief.com/en/article/claude-code-introduces-loop-engineering-via-goal-and-loop-32519 AX BRIEF en 2026-08-15T02:23:57.721Z Claude Code Introduces Loop Engineering via /goal and /loop https://axbrief.com/article/hubble-md-turns-markdown-notes-into-collaborative-agent-workspaces-32516 AX BRIEF ko 2026-08-15T02:10:06.491Z hubble.md, 사람-에이전트 협업용 오픈소스 노트 앱 공개 https://axbrief.com/en/article/hubble-md-turns-markdown-notes-into-collaborative-agent-workspaces-32516 AX BRIEF en 2026-08-15T02:10:06.491Z hubble.md Turns Markdown Notes Into Collaborative Agent Workspaces https://axbrief.com/article/deepseek-v4-pro-adds-reasoning-control-and-off-peak-api-pricing-49296627 AX BRIEF ko 2026-08-15T00:44:28.359Z DeepSeek, V4-Pro 공개 및 API 시간대별 차등 요금제 도입 https://axbrief.com/en/article/deepseek-v4-pro-adds-reasoning-control-and-off-peak-api-pricing-49296627 AX BRIEF en 2026-08-15T00:44:28.359Z DeepSeek V4-Pro Adds Reasoning Control and Off-Peak API Pricing https://axbrief.com/column/deepseek-v4-pro-weave-toolkit-and-pixel-11-accessibility-updates-260815 AX BRIEF ko 2026-08-15T00:07:15.127Z DeepSeek V4 Pro 성능 개선과 AI 개발 도구의 진화 https://axbrief.com/en/column/deepseek-v4-pro-weave-toolkit-and-pixel-11-accessibility-updates-260815 AX BRIEF en 2026-08-15T00:07:15.127Z DeepSeek V4 Pro, Weave Toolkit, and Pixel 11 Accessibility Updates https://axbrief.com/article/how-claude-code-uses-prompt-caching-to-optimize-token-costs-49300800 AX BRIEF ko 2026-08-14T22:47:55.529Z Claude Code, 프롬프트 캐싱 기반 토큰 비용 구조와 컨텍스트 최적화 방식 공개 https://axbrief.com/en/article/how-claude-code-uses-prompt-caching-to-optimize-token-costs-49300800 AX BRIEF en 2026-08-14T22:47:55.529Z How Claude Code Uses Prompt Caching to Optimize Token Costs https://axbrief.com/article/why-google-heir-is-the-missing-link-for-private-ai-inference-49300314 AX BRIEF ko 2026-08-14T20:52:22.186Z 암호화된 데이터 그대로 AI 추론하는 구글 HEIR 공개 https://axbrief.com/en/article/why-google-heir-is-the-missing-link-for-private-ai-inference-49300314 AX BRIEF en 2026-08-14T20:52:22.186Z Why Google HEIR is the Missing Link for Private AI Inference https://axbrief.com/article/why-did-qwen-ship-3-8-27b-with-a-reasoning-effort-parameter-49299605 AX BRIEF ko 2026-08-14T18:53:05.522Z "생각하고 답한다" Qwen 3.8-27B, 추론 제어와 26만 토큰 컨텍스트 공개 https://axbrief.com/en/article/why-did-qwen-ship-3-8-27b-with-a-reasoning-effort-parameter-49299605 AX BRIEF en 2026-08-14T18:53:05.522Z Why Did Qwen Ship 3.8-27B With a Reasoning Effort Parameter? https://axbrief.com/article/hexis-brings-gitops-to-ai-agent-skill-and-permission-management-49300784 AX BRIEF ko 2026-08-14T17:00:50.182Z Hexis, Git 저장소 기반 AI 에이전트 스킬 및 권한 제어 평면 공개 https://axbrief.com/en/article/hexis-brings-gitops-to-ai-agent-skill-and-permission-management-49300784 AX BRIEF en 2026-08-14T17:00:50.182Z Hexis Brings GitOps to AI Agent Skill and Permission Management https://axbrief.com/article/the-20-billion-collapse-of-leopold-aschenbrenners-situational-awareness-49299282 AX BRIEF ko 2026-08-14T16:50:03.566Z 레오폴드 아셴브레너의 헤지펀드 붕괴와 AI 랩의 '지적 오만' 리스크 https://axbrief.com/en/article/the-20-billion-collapse-of-leopold-aschenbrenners-situational-awareness-49299282 AX BRIEF en 2026-08-14T16:50:03.566Z The $20 Billion Collapse of Leopold Aschenbrenner's Situational Awareness https://axbrief.com/blog/why-amazon-nova-forge-shifts-to-grpo-for-multi-turn-rft-gx8q2dn AX BRIEF ko 2026-08-15T01:18:28.651Z Amazon Nova Forge의 멀티턴 RFT 구현과 보상 함수 설계 시 주의할 검증 경계 https://axbrief.com/en/blog/why-amazon-nova-forge-shifts-to-grpo-for-multi-turn-rft-gx8q2dn AX BRIEF en 2026-08-15T01:18:28.651Z Why Amazon Nova Forge Shifts to GRPO for Multi-Turn RFT https://axbrief.com/blog/chinese-2-78t-parameter-models-vs-us-optimization-the-new-ai-power-e8y4ch1 AX BRIEF ko 2026-08-14T23:03:04.825Z 2.78조 파라미터 중국 모델 vs 하드웨어 최적화 미국 모델, 오픈 소스 주도권의 변화 https://axbrief.com/en/blog/chinese-2-78t-parameter-models-vs-us-optimization-the-new-ai-power-e8y4ch1 AX BRIEF en 2026-08-14T23:03:04.825Z Chinese 2.78T Parameter Models vs US Optimization: The New AI Power Struggle https://axbrief.com/blog/how-claudes-new-synthid-text-watermark-avoids-degrading-model-quality-cw1kkml AX BRIEF ko 2026-08-14T20:37:31.745Z "품질 영향 없다", Anthropic이 공개한 Claude 텍스트 워터마크 작동 방식 https://axbrief.com/en/blog/how-claudes-new-synthid-text-watermark-avoids-degrading-model-quality-cw1kkml AX BRIEF en 2026-08-14T20:37:31.745Z How Claude's New SynthID-Text Watermark Avoids Degrading Model Quality https://axbrief.com/blog/nvaitc-how-nvidia-and-indosat-are-scaling-sovereign-ai-in-indonesia-avj9k4l AX BRIEF ko 2026-08-14T18:23:25.004Z 인도네시아가 NVIDIA·Indosat과 구축한 소버린 GPU 기반 AI 인재 양성 모델 https://axbrief.com/en/blog/nvaitc-how-nvidia-and-indosat-are-scaling-sovereign-ai-in-indonesia-avj9k4l AX BRIEF en 2026-08-14T18:23:25.004Z NVAITC: How NVIDIA and Indosat are Scaling Sovereign AI in Indonesia https://axbrief.com/blog/why-bedrock-agentcore-needs-sagemaker-ai-for-cost-optimized-multi-agents-ng12xth AX BRIEF ko 2026-08-14T17:46:21.030Z Bedrock과 SageMaker를 섞어 쓰는 비용 최적화 멀티 에이전트 설계법 https://axbrief.com/en/blog/why-bedrock-agentcore-needs-sagemaker-ai-for-cost-optimized-multi-agents-ng12xth AX BRIEF en 2026-08-14T17:46:21.030Z Why Bedrock AgentCore Needs SageMaker AI for Cost-Optimized Multi-Agents