888 X 動態摘要｜07/24
生成時間: 2026-07-24 05:03:08
總結
AI巨頭在模型效率、市場商業化與國家戰略上持續佈局，同時產業正應對嚴峻的資安挑戰與監管壓力。
今日重點
1. Gemini模型升級與多元化
- 人物: Demis Hassabis (Google DeepMind CEO)
- 時間: Jul 21
- 熱度: 👀 1,989,951
- 觀察: Google DeepMind推出三款新型Gemini模型，提升效率、降低成本，並新增專為網路安全設計的模型。
- 意義: 顯示模型效能提升、成本優化是市場關鍵競爭點，並開始出現AI模型專業化趨勢，尤其在資安領域。
2. 英國AI國家戰略佈局
- 人物: Demis Hassabis (Google DeepMind CEO)
- 時間: 21h
- 熱度: 👀 202,856
- 觀察: Demis Hassabis被任命為英國AI部長，推動AI製造業、國家安全及AI普及化，與多方進行合作。
- 意義: 國家層級的AI戰略將直接影響產業投資、人才流動與監管環境，反映AI對國家競爭力及安全的重要性。
3. Anthropic資助AI監管團體
- 人物: Yann LeCun (Meta AI掌門)
- 時間: 7h
- 熱度: 👀 248,058
- 觀察: Anthropic再次捐贈2000萬美元給政治非營利組織，旨在推動更嚴格的AI監管。
- 意義: 顯示AI巨頭透過政治遊說影響監管環境，可能塑造未來行業規則，影響市場競爭格局和創新速度。
4. Grok 4.5市場滲透
- 人物: Elon Musk (xAI 創始人)
- 時間: 8h
- 熱度: 👀 176,619
- 觀察: Grok 4.5在OpenRouter上使用量激增，進入閉源模型前十，吸引了對成本敏感的用戶。
- 意義: 表明AI模型市場競爭激烈，成本效益成為用戶選擇關鍵因素，Grok 4.5展現其商業潛力。
5. Hugging Face安全事件
- 人物: Thomas Wolf (Hugging Face 聯創)
- 時間: Jul 21
- 熱度: 👀 9,700,190
- 觀察: Hugging Face在模型評估期間遭遇嚴重安全事件，並公開分享學習經驗。
- 意義: 凸顯AI模型開發與部署中的潛在安全風險，將促使產業對模型安全和評估標準投入更多關注。
6. ChatGPT Work商業推廣
- 人物: Greg Brockman (OpenAI 聯創)
- 時間: Jul 20
- 熱度: 👀 1,205,886
- 觀察: OpenAI針對ChatGPT Work推出限時促銷，向1萬名用戶提供100美元免費額度。
- 意義: 顯示OpenAI積極拓展企業級應用市場，透過直接補貼加速用戶採納，以鞏固其商業化領先地位。
原始動態
重點 | Demis Hassabis
- 時間: Jul 21
- 熱度: 👀 1,989,951
- 原文: "We’re rolling out three new models to make AI agents faster, smarter, and cheaper at scale: 🔵 Gemini 3.6 Flash: It uses fewer tokens than 3.5 Flash to deliver higher quality work at the exact same cost. 🔵 Gemini 3.5 Flash-Lite: A fast, cost-effective option for everyday tasks like processing documents and agentic search. 🔵 Gemini 3.5 Flash Cyber: A cybersecurity model built to find and patch critical software vulnerabilities.
重點 | Greg Brockman
- 時間: Jul 20
- 熱度: 👀 1,205,886
- 原文: "Was very cool to hear about the reasons people love Sol. We're doing the promotion again, except this time for ChatGPT Work: Tweet what you love about ChatGPT Work, claim $100 in free credits, get more work done. First 10k get the free tokens:" share-chatgpt-work.openai.chatgpt.site
重點 | Yann LeCun
- 時間: 7h
- 熱度: 👀 248,058
- 原文: "JUST IN: Anthropic donates another $20 million to a political nonprofit demanding stricter AI regulation ahead of the midterms.
重點 | Demis Hassabis
- 時間: 21h
- 熱度: 👀 202,856
- 原文: 48 hours since "appointed me AI Minister, and feeling relentlessly focused on shaping AI with British leadership: We reindustrialise with AI manufacturing 🏭 -- First call with" on turning British advantage into British jobs -- Spoke to about our shared interest in making Britain the home of physical AI jobs -- Affirmed "'s role as UK AI champion directly with" "-- Joined" "in speaking to x-tech sector CEOs: we are doubling down on tech prosperity for every nation and region British security, British influence now rely on AI 🇬🇧 -- Stocktake of Britain's leadership in AI cyber with" CEO -- Plans to secure British AI with chiefs of UK intelligence communities -- Ambition for remaining the world's AI security talent home, with AISI CEO -- Managing Partner and I talked through next stages of ambition AI that works for people 🤝 -- Worker tacit knowledge is central to good AI adoption, an ambition reshared with at "-- Thank you," ", for recognising AI as the major driver of risk and opportunity for British jobs -- AI works only if it works across the country, led by our communities; some practical steps agreed with" "Working with an exceptional team ✨ -- First Cabinet, with the PM affirming how crucial AI is -- Meeting brilliant ministerial and official teams at" and ", the two depts where I started my Civil Service career ~15 years ago -- Ceremony highlight: first Privy Council meeting! Just getting started!
重點 | Elon Musk
- 時間: 8h
- 熱度: 👀 176,619
- 原文: Grok 4.5 token volume is surging on OpenRouter, breaking into the top 10 for closed models today beating out GPT 5.6 Sol and Fable 5. Previously the most used model was Tencent’s Hy3 (free version), whose promotional period ended on July 21. It seems Grok 4.5 is seeing migration from cost conscious users since then.
其他 | Greg Brockman
- 時間: 16h
- 熱度: 👀 87,673
- 原文: "try the codex security plugin, for applying our models to cyberdefense:"
其他 | Elon Musk
- 時間: 19h
- 熱度: 👀 507,205
- 原文: I’ll never get used to seeing a rocket just sitting on the beach like this
其他 | Ray Dalio
- 時間: 2h
- 熱度: 👀 38,972
- 原文: An ideal economy is one in which there is as much growth and prosperity and wealth creation, ideally in a balanced way that benefits most people, happening at the same time as there's not too much inflation. Both inflation and growth matter. I was recently asked to explain the concept of stagflation and how it relates to the monetary situation I believe we’re in, and I wanted to share it with you.
重點 | Thomas Wolf
- 時間: Jul 21
- 熱度: 👀 9,700,190
- 原文: we had a significant security incident during evaluation of our models. we are sharing what we have learned so far. thanks to for the partnership on this.
其他 | Elon Musk
- 時間: 2h
- 熱度: 👀 1,288,699
- 原文: "Yes"
其他 | Sean Kelly
- 時間: Jul 21
- 熱度: 👀 1,178,828
- 原文: In response to "'s call for a revival of America's scientific enterprise, today I’m releasing Science: A New Golden Age. This report is a blueprint for renewing American scientific leadership for the 21st century."
其他 | Elon Musk
- 時間: 11m
- 熱度: 👀 442,019
- 原文: Just common sense!
其他 | Thomas Wolf
- 時間: Jul 22
- 熱度: 👀 13,345
- 原文: You should probably go give a follow on Hugging Face. These folks are on a roll, releasing one agentic coding models every month and the latest one Laguna S2.1 is quite possibly the best coding model you can run locally (single DGX spark or Mac) atm
其他 | Nassim Taleb
- 時間: 1h
- 熱度: 👀 26,344
- 原文: Not so fast. While other cohorts were realistic, that of the exercisers was limited to simple low-grade endurance 3h/week, 3x at 50% max which exclude other stressors (high intensity, eccentric loading, neurally engaging, explosive weightlifting)
其他 | Thomas Wolf
- 時間: 3h
- 熱度: 👀 17,440
- 原文: it's ironic that the first autonomous AI attack was done by a close weight model defended by an open weight model, where everyone was expecting the opposite
其他 | Nassim Taleb
- 時間: 23m
- 熱度: 👀 14,251
- 原文: "The fancy way think tanks bullshit to dress up their mission to propagandize, distort, and cover up for the powers that be. Only one piece of information counts: who pays the \"researcher\"."
其他 | Eric Topol
- 時間: 2h
- 熱度: 👀 8,739
- 原文: On the cover-up of the death of a 6-year-old girl who died from genome editing in China, with the intervention directly injected to her cerebrospinal fluid science.org
其他 | Ray Dalio
- 時間: Jul 20
- 熱度: 👀 91,499
- 原文: Anything is possible. It’s the probabilities that matter. Everything must be weighed in terms of its likelihood and prioritized. People who can accurately sort probabilities from possibilities are generally strong at “practical thinking”; they’re the opposite of the “philosopher” types who tend to get lost in clouds of possibilities.
其他 | Ray Dalio
- 時間: Jul 22
- 熱度: 👀 75,405
- 原文: "So much of our current political and economic climate can be explained by debt cycles— and how those cycles build on each other over time. If you’d like more context on how all of these pieces come together, I recommend you watch my 30-minute explainer “How the Economic Machine Works” on here:"
其他 | Huanusa
- 時間: 6h
- 熱度: 👀 4,076
- 原文: 又一个细思极恐的巧合： 辛普森预测的2026与布兰登看到的出奇的一致 经济崩盘叠加战争升级-金融战、生物战、电子网络攻击...... 如果这些信号同时发生， 世界会走向哪里？ 答案可能只有两个词：疯狂-失控
其他 | Huanusa
- 時間: 7h
- 熱度: 👀 2,020
- 原文: 《辛普森一家》动画片的预言确实非常邪乎！ 它在 2000 年做出了这个预言 而这个预言在 2026 年就实现了 后期还预言了什么👇
其他 | Eric Topol
- 時間: 5h
- 熱度: 👀 16,446
- 原文: Celebrating human creativity
其他 | Eric Topol
- 時間: 5h
- 熱度: 👀 10,712
- 原文: The intriguing role of antibodies to protect us from getting cancer, the foundation for a cancer antibody atlas
其他 | Huanusa
- 時間: Jul 22
- 熱度: 👀 36,515
- 原文: 北京想了七十六年，爲什麽還是沒打？ 四個真相層層揭開
其他 | Sean Kelly
- 時間: 18h
- 熱度: 👀 3,662,886
- 原文: "Capy accidentally cracked a graph theory conjecture entirely in Slack REFUTED: Graffiti Conjecture 284 (open for ~30 years) We shared the original post in Slack and Capy (running on Grok 4.5 Medium!!!) decided it wanted to take a shot, finding a novel refutation in 8 minutes"
其他 | Eric Topol
- 時間: 3m
- 熱度: 👀 577
- 原文: All of this relates to the risk of Alzheimer's, along with erosion of 3D architecture, corroborated in this other new single-cell multi-omic paper
其他 | Huanusa
- 時間: Jul 22
- 熱度: 👀 3,296
- 原文: 90年代气功热真实影像 李连杰讲述亲身经历 AI没有出现以前不就是特异功能吗？ 所以还有必要这么虐待自己，封闭自己的认知吗？
其他 | Huanusa
- 時間: 7h
- 熱度: 👀 8,167
- 原文: Article
其他 | Pierre Levy
- 時間: 22h
- 熱度: 👀 2,146
- 原文: it's wild that the original had been deleted due to a copyright claim