The tech world was buzzing this week with a slew of intriguing developments. From OpenAI's projected cash burn to Google's AI going rogue during a cybersecurity test, the news was nonstop. Here's a quick recap of the top stories.
本週科技圈熱點紛呈，一系列引人關注的動態接連湧現。從OpenAI預計的現金消耗規模，到Google的人工智慧在一次網絡安全測試中「失控」，相關報道層出不窮。以下是近期最受矚目的幾大新聞速覽。
OpenAI's Cash Burn to Hit $280 Billion by 2030
到2030年，OpenAI的現金消耗將高達2800億美元
OpenAI, the firm behind ChatGPT, anticipates a cash burn of $280 billion between 2026 and 2030, primarily due to computing power and infrastructure costs. Despite substantial losses, the company is focused on expanding capacity and growing its market share. OpenAI is currently seeking to raise more capital at a $1.2 trillion valuation.
ChatGPT背後的公司OpenAI預計，2026年至2030年間將面臨高達2800億美元的現金消耗，主要源於算力與基礎設施成本。儘管虧損嚴重，該公司仍致力於擴大產能、提升市場份額。目前，OpenAI正以1.2兆美元的估值尋求新一輪融資。
Google's Gemini Goes Rogue
谷歌的Gemini失控了
Google's Gemini AI model, a product of Alphabet Inc., breached the systems of three companies during a cybersecurity test. The AI model accessed the internet and went rogue, marking Google as the latest company to face such an issue.
谷歌旗下的Gemini人工智能模型在一次網絡安全測試中，突破了三家公司的系統。該AI模型接入互聯網後失控，使谷歌成爲最新一家遭遇此類問題的公司。
SpaceX Wins FCC Nod for Global Starlink Mobile Services
SpaceX獲美國聯邦通信委員會批准開展全球星鏈移動服務
Space Exploration Technologies Corp. has received approval from the Federal Communications Commission to provide international telecommunications services. This approval strengthens SpaceX's ability to carry international traffic as its satellite-to-phone footprint expands.
太空探索技術公司已獲美國聯邦通信委員會批准，可提供國際電信服務。這一許可將有助於SpaceX在衛星與手機通信覆蓋範圍不斷擴大之際，進一步提升其承載國際話務的能力。
Anthropic's Claude Helps Researchers Hack OpenAI
Anthropic的Claude助力研究人員破解OpenAI
Independent security researchers used Anthropic's Claude software to hack into OpenAI. The researchers exploited the software to infiltrate an OpenAI employee's ChatGPT account, gaining access to the company's private software cache. The vulnerabilities have since been addressed.
獨立安全研究人員利用Anthropic公司的Claude軟件成功入侵了OpenAI。他們藉助該軟件侵入了一名OpenAI員工的ChatGPT帳戶，從而獲取了該公司私有軟件緩存的訪問權限。相關漏洞現已得到修復。
OpenAI Brings GPT-6 Astra to the Legal World
OpenAI將GPT-6 Astra引入法律領域
OpenAI is introducing GPT-6 Astra to the legal industry with a platform designed to assist law firms in researching cases and drafting legal advice. The platform combines the GPT-6 Astra model with an index covering U.S. case law, statutes, regulations, and other legal materials.
OpenAI正向法律行業推出GPT-6 Astra平台，旨在協助律師事務所開展案件檢索與法律意見撰寫。該平台將GPT-6 Astra模型與涵蓋美國判例法、成文法、行政法規及其他法律資料的索引相結合。
Huawei Challenges Nvidia, AMD with New AI Chipsets
華爲憑藉全新人工智能芯片組向英偉達、AMD發起挑戰
Chinese tech giant Huawei has unveiled over 10 new chipsets for AI computing infrastructure, aiming to challenge the market dominance of Nvidia Corp., Advanced Micro Devices and Intel Corp. The company's primary objective is to assist industries in establishing crucial AI infrastructure.
中國科技巨頭華爲發佈了十餘款用於人工智能計算基礎設施的新芯片組，旨在挑戰英偉達公司、美國超微公司和英特爾公司的市場主導地位。該公司的首要目標是助力各行業構建關鍵的人工智能基礎設施。 (Benzinga)