5분 만에 오늘의 테크 파악

AI가 매일 HN·Reddit·주요 테크 미디어의 150개+ 중 읽을 가치 있는 30개를 선별합니다. 구독하면 당신의 관심사에 맞춰 큐레이션하여, 요약과 커뮤니티 반응과 함께 매일 아침 당신만의 다이제스트를 보내드립니다.

매일 아침 6:30 발송 · 언제든 원클릭 해지

89호 발행 · 매일 150개+ 중 읽을 가치 있는 30개로 선별

매일 주요 소스를 선별

Hacker NewsRedditGitHubarXivHugging FaceThe VergeArs TechnicaWired

dawnsift.com/d/2026-10-09

금 · 테크 데일리 · 제89호

2026-10-09

오늘의 TL;DR

Google releases Gemini Agent, formally entering the office agent melee, with support for calling multiple models such as Claude. JetBrains open-sources the 12B MoE coding model Mellum2.1, with SWE-bench Verified jumping from 2.0 to 47.0. Anthropic launches the free open-source security scanning service OSS Scanner. On security, the South Korean bank attack incident shows a single person can combine open-source penetration tools and multiple LLMs to complete an intrusion.

Google releases Gemini Agent: a general office agent that supports calling Claude

Last week some of South Korea's biggest banks were hit by a cyberattack. We now know the entire hack may have been done by a single person. He used a combined stack of an open-source AI penetration tool named ARTEX, DeepSeek v4.1-Flash, GLM-5.3, Grok 4.6, and Claude Code (CrowdStrike)reddit · 1 min

Anthropic launches free AI security scans for open-source projectsThe Verge · 1 min

RunningTab: Direct Workspace Interaction with Environment-Side Tabshf_papers · 2 min

전문 보기 →