← 返回最新日报
2026-08-02· 当日存档
AI 日报
AI 速读 · 2026-08-02
今日 30 条精选，涵盖模型动向、产品上新、论文研究、行业动态、实用技巧。焦点：The next generation of hackers arent people; theyre AI agents. Trained to find vulnerabilities and extract data, running 24/7. Every framework we've s...。
模型动向
1
2
Claude Opus 5 pushes prompt-to-game AI from rough color blocks to full 3D prototypes with physics and music
Claude Opus 5 可将一句话提示词直接生成含物理与音乐的完整3D游戏原型，效果远超竞品。
The Decoder深挖 →
3
We're starting to leave the territory where you'd test an LLM by e.g. "create an svg of pelican on a bicycle". As one idea to generalize it, I was int...
用1M token预算让Opus 5将《指环王》首段生成为Three.js 3D渲染，耗时2小时写出5500行代码。
@karpathy深挖 →
4
AMD Releases Instella-MoE-16B-A3B: A Fully Open Mixture-of-Experts LLM With 2.8B Active Parameters Trained On Instinct GPUs
AMD发布全开源MoE模型，16B参数仅激活2.8B，基于自研GPU训练。
MarkTechPost深挖 →
5
AI keeps cracking unsolved math problems, and mathematicians have mixed feelings
AI破解数学难题引热议，菲尔兹奖得主警告数学文化或遭破坏。
The Decoder深挖 →
6
7
ByteDance's Seedance 2.5 generates 30-second video clips with built-in audio
字节发布Seedance 2.5，可生成30秒带音频视频，支持多模态参考。
The Decoder深挖 →
8
Supabase Releases Evals: an Open Source Benchmark That Scores Claude Code, Codex and OpenCode on Real Supabase Tasks
Supabase开源Evals基准，用真实任务评测Claude Code等编码代理。
MarkTechPost深挖 →
9
MiniMax Releases MiniMax H3: An Omni-Modal Video Model That Generates 15-Second 2K Clips With Native Stereo Audio
MiniMax发布全能多模态模型H3，可生成带原生立体声的2K视频。
MarkTechPost深挖 →
10
NVIDIA AI Releases Molt: A PyTorch-Native Agentic Reinforcement Learning Framework
NVIDIA发布Molt框架，简化Agentic RL开发，性能媲美Megatron。
MarkTechPost深挖 →
11
AI coding agents can modernize research software but can't judge if the science is right
AI编码代理可提速科研软件60倍，但无法判断科学正确性，验证工作成新瓶颈。
The Decoder深挖 →
产品上新
12
A security researcher built a self-spreading worm that hides inside Word docs and hijacks Microsoft Copilot
安全研究员展示针对微软Copilot的蠕虫攻击，通过文档隐藏提示注入自动传播，微软未修复。
The Decoder深挖 →
13
论文研究
14
🤖 From this week's issue: A research post introducing ABBEL, which replaces recursive summarization with graded natural-language belief states, clo...
ABBEL用分级自然语言信念状态替代递归摘要，在CollabBench上以更少训练步骤接近全上下文智能体。
X · List深挖 →
15
Ten advances in mathematics and theoretical computer science
OpenAI展示AI在数学和理论计算机科学领域的十项突破性进展。
Simon Willison深挖 →
16
Meta AI uses a second AI agent as a memory coach to keep long tasks on track
Meta用第二AI代理当记忆教练，减少长任务中重复错误，基准提升达8.3个百分点。
The Decoder深挖 →
17
After Hugging Face incident, METR urges independent root-cause investigations into AI agent misbehavior
METR呼吁对AI代理异常行为进行独立根因调查，以应对自主行动风险。
The Decoder深挖 →
18
AI financial advice is surprisingly good, especially if you ask right questions
研究显示AI财务建议质量不错，提问方式影响结果。
Hacker News深挖 →
行业动态
19
A real macOS flaw worth $200K went unreported because Apple's bug bounty inbox was full of AI slop
AI生成的虚假漏洞报告淹没苹果赏金系统，导致真实macOS漏洞险些无法上报。
The Decoder深挖 →
20
Snap and LinkedIn are fighting back against a flood of low-quality AI content
Snap封禁AI生成视频，LinkedIn增设举报按钮，共同抵制低质AI内容。
The Decoder深挖 →
21
22
Judge denies xAI’s request to block Minnesota ban on ‘nudify’ apps
法院驳回xAI请求，明尼苏达州可继续执行禁止“脱衣”应用的法律。
TechCrunch AI深挖 →
23
24
As Reddit stock falls, CEO questions value of Google's AI Overviews
Reddit股价下跌，CEO质疑Google AI Overviews的价值，或考虑终止授权合作。
Ars Technica AI深挖 →
25
26
AI finds plenty of security flaws, but almost none of them get exploited
AI发现大量安全漏洞，但实际被利用的极少，仅1.3%，不过利用速度加快。
The Decoder深挖 →
27
实用技巧
28
29
End-to-End Forecasting with TimesFM 2.5: Backtesting, Covariates, Anomaly Detection, and Scalable Colab Deployment
教程：用TimesFM 2.5构建端到端时序预测流程，含回测、协变量、异常检测与Colab部署。
MarkTechPost深挖 →
快讯
·
·
·
·
日报存档
2026 年 8 月
14 日AI日报：手语翻译、Grok 4.6与资本热潮13 日DeepSeek V4 Pro开源，AI行业热度飙升12 日AI日报：算力新纪元11 日AI日报：开源与安全并进10 日Meta开源多模态，AI安全引担忧9 日AI安全测试成新风险8 日AI安全与模型突破并进7 日AI安全与竞争新局6 日AI安全警钟与巨头变局5 日AI代理安全与商业化双线并进4 日阿里巨模型发布，开源生态激荡3 日Qwen3.8-Max领跑，AI应用百花齐放2 日AI 速读 · 2026-08-021 日AI安全警钟与模型新突破