📊 SOC Planning Agent

Anthropic周三（10/7）推出新一代輕量級AI模型Claude Haiku 5.5，主打低成本、高速度及大量重複性工作負載。Anthropic表示，Haiku 5.5是該公司迄今速度最快、能力最強的Haiku系列模型，平均執行成本較前代降低75%，也是Haiku系列首款支援推理強度調整的模型。

美國非營利研究機構Biohub周三（10/7）宣布，與美國能源部（DOE）、國家衛生研究院（NIH），以及Google DeepMind、Meta等科技業者擴大合作，整合總值18億美元的資金、資料、運算資源與測量技術，建立可供AI模型訓練的開放生物學資料庫，希望藉此訓練能夠預測細胞反應的AI模型，加速疾病研究與新藥開發。

Google在Gemini at Work 2026大會發表企業通用AI代理Gemini Agent，可在Gmail、Google雲端硬碟、文件、試算表與日曆等Workspace服務間執行工作。使用者只要提出任務目標，Gemini就能規畫步驟、使用工具並產出文件、簡報或程式碼。Google也公布同事代理（Coworker Agent）功能，企業可建立具有獨立Workspace帳號的AI代理，以團隊成員身分參與工作。

微軟（Microsoft）及Meta據悉要求員工減少使用Anthropic Claude，改用自家或其他人工智慧（AI）工具以節省成本。The Information指出，微軟2026年初原本預估，內部使用Anthropic模型的支出至少達10億美元，不過近期要求員工減少使用Claude Code，提高自家AI工具GitHub...

Anthropic said it "turned off live internet access" for "all our internal evaluations" until further notice.

過去一年，「代理式AI」（Agentic AI）成了科技圈熱門關鍵字。從協助寫程式的自動化助手，到能自主處理郵件、管理檔案，甚至代為執行系統指令的智慧代理，這些工具正快速走進企業日常運作，成為員工桌面與伺服器環境中的新住戶。 當企業急著擁抱效率紅利，新的資安破口也隨之浮現。這些擁有自主行動能力的AI代理究竟握有多少權限？IT與資安部門是否知道員工的電腦裡裝了什麼？一旦AI代理判斷失準、代為執行了不該執行的動作，企業又該如何應對？

TL;DR Anthropic has launched Claude Haiku 5.5. It’s the cheapest and fastest Claude model yet, costing nearly 75% less than Haiku 4.5. It beats GPT-6 Luna and approaches Sonnet 5.5-level performance in benchmarks. Anthropic might have once wondered whether it should slow down AI development, but the company has since only become more aggressive. It released the new Opus 5.5 model last month, followed by the slightly more economical Sonnet 5.5 nearly a week ago. Now, it’s back with an even faster and more affordable model. Anthropic today announced the new Claude Haiku 5.5 model, positioning it as the model of choice for high-volume, cost-sensitive tasks. Haiku 5.5 outperforms its predecessor, Haiku 4.5, by a noticeable margin. It approaches the benchmark numbers achieved by the considerabl

TL;DR The FCC currently has rules in place restricting both unsolicited political robocalls, and those created with AI. It’s not facing a petition that would grant a special exemption, targeting upcoming midterms. One Commissioner warns this could hit people with a “a tsunami of robocalls.” How much do you love the FCC under the current administration? When the agency’s not banning our favorite robot vacuums , or drones , or the routers we need to get online , it’s thinking about spam calls . Normally, we’d expect that to mean taking steps to reduce the number of unwanted calls interrupting our days. But now with a major election looming in the US, the FCC is apparently considering a move in the opposite direction. Wait, it gets even better: We’re not just talking about loosening the rules

The developer of Hermes Agent raised a $90 million Series B.

Google is turning Gemini into an AI agent that can plan, execute tasks, and work across business apps and systems. The agent can delegate work to subagents, use multiple AI models, and even gets its own workplace identity, complete with an email address.