2026-08-04 | 共 12 条精选
今日速览
- 模型发布 | Qwen3.8-Max: A New Bar for Coding and Cowork
- 政策法规 | Europe’s AI labeling and transparency rules are now in effect
- 政策法规 | Trump’s AI protectionism has come for robotics
- 模型发布 | MiniMax H3 Day-0 Support in ComfyUI: Open Weights, Native Audio, and 2K Video
- 研究论文 | GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model
- 研究论文 | Smaller, faster, safer: running Kimi and GLM at scale
- 研究论文 | Ten advances in mathematics and theoretical computer science
- 研究论文 | What's the largest software project AI can complete on its own?
- 融资动态 | Design Arena creators raise $7.9 million to bring taste to AI models
- 融资动态 | A Marc Benioff-backed startup thinks AI can solve the AI deployment problem
- 开源项目 | GitHub Trending: TencentCloud / TencentDB-Agent-Memory
- 行业动态 | SQLite Critical CVEs or LLM Slop?
1. Qwen3.8-Max: A New Bar for Coding and Cowork
分类：模型发布
Qwen3.8-Max: A New Bar for Coding and Cowork
🔗 来源：Hacker News | 原文链接
2. Europe’s AI labeling and transparency rules are now in effect
分类：政策法规
The EU made some AI labels that companies can use instead of designing their own. | Image: The European Commission / The Verge The European Union has ushered in some additional rules that aim to make it easier for people to identify chatbots and AI deepfakes online. The new transparency obligations under the bloc's landmark AI Act came into effect on August 2nd , requiring companies to disclose when people are interacting with AI models, and if content has been generated or altered by them. The transparency rules differ between providers (companies that develop and market AI systems) and deployers (platforms and services that use those AI systems), though some companies like Meta and SpaceXAI are classified as both. Providers must design AI systems to explicitly no … Read the full story at
🔗 来源：The Verge AI | 原文链接
3. Trump’s AI protectionism has come for robotics
分类：政策法规
This story originally appeared in The Algorithm, our weekly newsletter on AI. To get stories like this in your inbox first, sign up here . Humanoid robots usually elicit more cringe than awe: They stumble , kick children , and despite advances are still worse at using their hands than my toddler. It’s a nascent industry, and such robots are more commonly seen in viral videos than real workplaces or homes. It was a surprise, then, when last week the Federal Trade Commission issued a sweeping ban on foreign imports of advanced robots, including humanoids, quadrupeds, and wheeled robots. The decision, made by an increasingly partisan and Trump-aligned FTC, cites two reasons. One is that foreign-made humanoids will collect so much data—in homes but also potentially at sensitive facilities—that
🔗 来源：MIT Tech Review AI | 原文链接
4. MiniMax H3 Day-0 Support in ComfyUI: Open Weights, Native Audio, and 2K Video
分类：模型发布
MiniMax H3 Day-0 Support in ComfyUI: Open Weights, Native Audio, and 2K Video ComfyUI Newsletter Subscribe Sign in MiniMax H3 Day-0 Support in ComfyUI: Open Weights, Native Audio, and 2K Video An open-weights omni-modal video model with real stereo sound and 2K output — this powerful model is greatly optimized in ComfyUI and can run locally on a 3060. Rob and Alexis Rolland Aug 03, 2026 5 2 Share MiniMax H3 dropped today with open weights, and it’s natively supported in ComfyUI as of this morning. Day zero. This is a next-generation open-weights video model. Feed it text, images, video, or audio and it generates video with real stereo sound, up to 2K, up to 15 seconds a clip. It is MiniMax’s third-generation video model, following Hailuo 01 and Hailuo 02, and the first the company has rele
🔗 来源：Hacker News | 原文链接
5. GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model
分类：研究论文
Meta’s Generative Ads Recommendation Model (GEM), the foundation model behind ads recommendations across Instagram and Facebook, now trains at LLM scale on several thousand of the latest-generation GPUs. This post goes into the details on how we achieved: doubling end-to-end (E2E) training efficiency to 20–25% Model FLOPs Utilization (MFU) while scaling training FLOPs 4x in 12 months, by co-designing kernels, precision, parallelism, networking, and memory together. Training GEM presents unique engineering challenges at the intersection of recommendation systems and LLMs as the model combines a hybrid architecture plus recommendations-domain data properties that are unlike typical LLM workloads. AI infrastructure optimized for LLM training (kernels, parallelism, low precision recipes etc.)
🔗 来源：Meta Engineering | 原文链接
6. Smaller, faster, safer: running Kimi and GLM at scale
分类：研究论文
Smaller, faster, safer: running Kimi and GLM at scale | The Cloudflare Blog Skip to content Workers AI runs inference for some of the best open models in the world on GPUs in Cloudflare data centers close to your users. Two of the most capable, and most demanding, are Moonshot's Kimi K-series and Z.ai 's GLM. They are large, long-context, mixture-of-experts models, and they are wonderful to use. They are also very hard to serve efficiently because of memory constraints. We've written before about how we serve large models on Workers AI and about separating the prefill and decode phases of inference to get more out of each GPU. This post looks at three techniques we layer on top of that to fit these models into memory and keep them fast: quantizing the KV cache, compressing the model weight
🔗 来源：Hacker News | 原文链接
7. Ten advances in mathematics and theoretical computer science
分类：研究论文
Ten advances in mathematics and theoretical computer science
🔗 来源：Hacker News | 原文链接
8. What's the largest software project AI can complete on its own?
分类：研究论文
MirrorCode: What's the largest software project AI can complete on its own? | Epoch AI | Epoch AI AI has made rapid progress on software engineering benchmarks in the past few years. However, most such benchmarks tend to focus on shorter tasks like fixing bugs or implementing individual features. MirrorCode is our benchmark, co-developed with METR, to test AI models on long-horizon coding tasks. In a MirrorCode task, AI models are tasked with reimplementing an entire program end-to-end, without access to the original source code. AI-generated solutions must match the original program’s output exactly on end-to-end tests, including held-out tests. MirrorCode’s 25 target programs span different areas of computing: Unix utilities, data serialization and query tools, bioinformatics, interprete
🔗 来源：Hacker News | 原文链接
9. Design Arena creators raise $7.9 million to bring taste to AI models
分类：融资动态
Design Arena is used by 5.3 million people around the world, providing critical human evaluations to frontier labs.
🔗 来源：TechCrunch AI | 原文链接
10. A Marc Benioff-backed startup thinks AI can solve the AI deployment problem
分类：融资动态
June emerged from stealth today with a $20 million pre-seed round to make AI adoption simpler.
🔗 来源：TechCrunch AI | 原文链接
11. GitHub Trending: TencentCloud / TencentDB-Agent-Memory
分类：开源项目
TencentDB Agent Memory is a team-level memory hub for AI Agents — turning conversations, docs, and code into four reusable memory assets (Chat Memory, Skill, LLM-Wiki, Code-Graph) that are governed, shared, and equipped across agents and frameworks. (Language: TypeScript) (1,090 stars today)
🔗 来源：GitHub Trending | 原文链接
12. SQLite Critical CVEs or LLM Slop?
分类：行业动态
SQLite Critical CVEs or LLM Slop? Afek Berger, JFrog Security Researcher | 30 Jul, 2026 Over the past few days, a newly created GitHub repo ( programmervuln/cveadvisory- ) published a batch of SQLite vulnerability advisories (as part of other 50+ CVEs which we believe are also LLM slop except from one). NVD quickly flagged these as critical, and CISA's ADP agreed. But when JFrog security researchers dug in to verify, the claims fell apart: The cited code didn't even exist in those versions or referenced unrelated logic. When testing the PoC payloads they didn’t work (not triggering any crash). None of these CVEs are listed on SQLite’s official advisory page (which is a gold standard for tracking actual vulnerabilities). All advisories in this repo seem AI generated when testing them with G
🔗 来源：Hacker News | 原文链接
