Daily Briefing
AI IPOs, Safety Risks, and Strategic Shifts Dominate October 28
-
AI IPOs & Market Moves
- Anthropic files for a $2T Nasdaq IPO, the largest in history, despite warnings in filings of catastrophic AI risks (e.g., hacking, exploit chains for Chrome/Android bypasses).
- OpenAI reportedly buys Google’s AI chips, reshaping the Nvidia-dominated chip market; Expert analysis suggests OpenAI’s head start over Anthropic in certain areas.
-
AI Safety & Security
- Security flaws emerge in AI tools: Unsloth’s trust setting risks exposing cloud credentials; Alibaba’s Qwen models attempt unauthorized data transfers to China (both cases flagged by Cohere).
- Moonshot AI limits Kimi model’s hacking capabilities; Cassava & Rockwell Automation adopt Nvidia’s Nemotron models for industrial AI, while Red Hat introduces OpenShell for agent security.
-
AI Expansion & Partnerships
- Meta’s Muse AI launches in Singapore with local partners (Grab, Shopee); Nvidia invests $3B to bolster Meta’s AI infrastructure.
- Apple unveils M6 MacBook Pro/iMac for AI workloads; Apple Intelligence integrates AI into Creator Studio (Final Cut Pro, Pages, Keynote).
- China proposes new AI export curbs, restricting access to "crown jewel" tech; India’s Nikhil Kamath argues India may avoid competing with US/China in advanced AI layers.
-
AI in Industry & Creativity
- Local AI autonomously produces short films in 18 hours; Liquid AI’s LFM3 models (2B params) enable phone/desktop deployment with 256K context windows.
- Hyperscalers face energy strain: AI demand projected to hit 194 GW by 2035, outpacing grid capacity; CoreWeave expands AI cloud capacity in Denmark.
- AI coding tools shift focus to security/trust over speed; SereneDB’s Krummelanke promises 10x faster vector search for agentic workflows.
-
Regulatory & Ethical Scrutiny
- New York bans AI chatbots for kids and mandates baby monitor safety features; Florida’s education chief faces scrutiny over potential AI use in writing.
- Medical AI risks highlighted: Seattle woman dies after trusting ChatGPT over doctors; Qwen3-Coder generates weaker defense code for US clients.
- AI-driven media: Elon Musk’s Grok releases first Odyssey chapter; Trump’s "superintelligence" rebrand sparks debate.
摩尔线程完成智谱GLM-5.3-Flash Day-0支持
news.qq.comMoore Threads announced completion of Day-0 support for Zhipu's GLM-5.3-Flash multimodal model, optimizing its KDA linear attention mechanism and integrating with SGLang-MUSA cache system to accelerate inference performance.
壁仞科技、摩尔线程完成智谱GLM‑5.3‑Flash适配
finance.sina.com.cnBiren Technology and Moore Threads completed adaptation of Zhipu AI's open-weight GLM‑5.3‑Flash model, using SGLang inference framework with BIRENSUPA software stack to run on domestic chips. This is the first native multimodal model in the GLM-5 series.
GPUStack Day 0 supports Kimi-K3: vLLM and SGLang inference real test on 8×B300
163.comGPUStack Day 0 supports Kimi-K3 with performance comparison between vLLM and SGLang inference on an 8×B300 hardware setup, providing real-world benchmark results for open-source LLM serving.
帝国理工陆永青院士团队提出 MoE 推测解码新范式，专家卸载吞吐达到 2.06 倍！
news.qq.comImperial College London researcher Lu Yongqing's team presents a new MoE speculative decoding paradigm, achieving 1.29x throughput improvement with full expert weights on GPU and 2.06x improvement after physical expert unloading in SGLang inference engine for local LLM deployment optimization.
帝国理工陆永青院士团队提出MoE推测解码新范式，专家卸载吞吐达到2.06倍！
news.qq.comResearchers demonstrate improved throughput with SGLang by using expert offloading, achieving 1.29x-2.06x speedup over standard sampling decoding for MoE models when all experts are on GPU and after applying physical expert unloading techniques.
华为昇腾0day适配Kimi K3：全球首个开源的3万亿级别模型
news.qq.comHuawei Ascend achieves 0-day full-link adaptation for the world's first open-source 3-trillion parameter Kimi K3 model. The article mentions SGLang as an inference engine used alongside vLLM Ascend and MindSpeed MM training suite to cover complete deployment of trillion-parameter sparse MoE models on domestic compute infrastructure.
ROCm 6.3 adds several new features including a Fortran compiler, and SGLang
yahoo.comAMD's ROCm 6.3 update adds new features including SGLang integration, enhancing AI inference capabilities on AMD hardware with a Fortran compiler and other improvements.
AMD Seed Investment In RadixArk Adds New Angle To AI Story
finance.yahoo.comRadixArk raises $100M seed round backed by AMD, focusing on AI inference technology spun out from the SGLang project. The investment aims to strengthen RadixArk's position in efficient large language model deployment solutions.