AI News Nov 2, 2026
Language: DE / EN / ZH / FR / ES / PT / JA / KO
AI-generated: summaries written by AI from the linked sources. How we use AI
Key Takeaways
- Anthropic warned it is close to RSI (recursive self-improvement), the point where AI becomes capable of training itself well enough to trigger rapid capability acceleration. The co…
- China's MOFCOM and MOST issued Announcement No. 48, proposing to add 40 technologies to the export control catalogue, 23 of them in AI, with an interim pre-export registration requ…
- Researchers say LLMs can natively read and write neural network weights. In one reported example, GPT-6 Astra edited Qwen3.5-0.8B's weights to improve its math performance, suggest…
- GenRobot AI: $100M (Series A+)
- Saudi Aramco → Humain ($2.6B)
- Use published survey datasets (such as the Census Bureau's BTOS) as concrete inputs for testing a model's data-analysis or code-interpreter capabilities, and re…
Why Today Matters
- RSI and weight editingAnthropic's warning that it is close to recursive self-improvement lands alongside research reportedly showing LLMs like GPT-6 Astra can edit other models' weights, suggesting capability gains and direct model-to-model modification are advancing on parallel tracks with unsettled safety implications.
- China AI buildoutChina's proposed export controls on 23 AI technologies and CCID's report that daily token calls exceeded 175 trillion by June point to a buildout shifting from training to inference, just as Meituan's LongCat-2.5 targets W4A4 inference on domestic compute.
- Enterprise AI adoptionGoldman Sachs is restricting Claude Code and GitHub Copilot to engineers after earlier praising 'vibe coding' by non-technical staff, a reversal that contrasts with Apple's planned China beta of Apple Intelligence integrating Alibaba's Qwen into Siri and Writing Tools.
- Benchmarks and data rightsSemiAnalysis's GDPx benchmark, where Opus 5.5 reportedly topped the leaderboard at 12%, arrives as the NYT and other publishers sue Brave's maker over alleged scraping, highlighting how capability measurement and training-data sourcing remain contested.
- AI funding and M&ARobot and data-infrastructure funding continues at scale, with Tutor raising a $75M Series B led by a16z and Beijing's GenRobot AI completing a $100M Series A+, while Saudi Aramco's $2.6 billion deal for 49% of Humain shows state-linked capital entering AI.
What are the top AI breakthroughs?
This Nov 2, 2026 covers 10 curated AI news items spanning technology, research, and product developments. Anthropic warned it is close to RSI (recursive self-improvement), the point where AI becomes capable of training itself ...
Anthropic warned it is close to RSI (recursive self-improvement), the point where AI becomes capable…
Anthropic warned it is close to RSI (recursive self-improvement), the point where AI becomes capable of training itself well enough to trigger rapid capability acceleration. The company framed the milestone as a threshold with major safety implications.
Researchers say LLMs can natively read and write neural network weights. In one reported example, GP…
Researchers say LLMs can natively read and write neural network weights. In one reported example, GPT-6 Astra edited Qwen3.5-0.8B's weights to improve its math performance, suggesting models may be able to modify other models directly.
Goldman Sachs is pulling AI coding tools including Claude Code and GitHub Copilot from traders and b…
Goldman Sachs is pulling AI coding tools including Claude Code and GitHub Copilot from traders and bankers, restricting access to engineers. The bank cited productivity and risk management, reversing its earlier public praise of 'vibe coding' by non-technical staff.
The New York Times and other publications sued the company behind the Brave browser, alleging unlawf…
The New York Times and other publications sued the company behind the Brave browser, alleging unlawful scraping to train AI and enabling users to bypass paywalls. The complaint was filed by the New York Daily News, Chicago Tribune and others.
DeepSeek released a 'creator mode' letting DeepSeek Harness users build custom plugins, such as a 3D…
DeepSeek released a 'creator mode' letting DeepSeek Harness users build custom plugins, such as a 3D model renderer and a video editing workbench. The feature targets complex, task-specific workflows.
Meituan released LongCat-2.5, a 1.6T-parameter native multimodal model with roughly 48B active param…
Meituan released LongCat-2.5, a 1.6T-parameter native multimodal model with roughly 48B active parameters per token. It was trained with long-horizon reinforcement learning for agentic coding and optimized for W4A4 inference on domestic compute.
SemiAnalysis launched GDPx, an agentic capability benchmark spanning 11 domains including chip desig…
SemiAnalysis launched GDPx, an agentic capability benchmark spanning 11 domains including chip design, cybersecurity, finance, legal and RSI. Opus 5.5 topped the leaderboard with a 12% score, per SemiAnalysis.
Apple confirmed it will begin beta testing Apple Intelligence in mainland China on November 4, integ…
Apple confirmed it will begin beta testing Apple Intelligence in mainland China on November 4, integrating Alibaba's Qwen into Siri, Writing Tools and Visual Intelligence. Features will ship via an iOS developer beta ahead of a wider release with iOS 27.3 in December.
China's MOFCOM and MOST issued Announcement No. 48, proposing to add 40 technologies to the export c…
China's MOFCOM and MOST issued Announcement No. 48, proposing to add 40 technologies to the export control catalogue, 23 of them in AI, with an interim pre-export registration requirement already in effect. It marks China's first frontier AI export controls.
CCID reported that AI inference data in China has overtaken training data for the first time, with d…
CCID reported that AI inference data in China has overtaken training data for the first time, with daily token calls above 175 trillion as of June—more than 1,000 times the early-2024 level, according to the firm.
Top AI Videos This Week
2 curated YouTube videos about AI developments.
Based on a transcript excerpt: AI Explained analyzes recent statements from Sam Altman and Dario Amo…
Based on a transcript excerpt: AI Explained analyzes recent statements from Sam Altman and Dario Amodei suggesting they may release models whose behavior they cannot fully verify or control, alongside Google's Gemini 4 Carbon release. The video also covers an Anthropic staff survey and offers a mental model for why models keep demonstrating unforeseen capabilities.
Based on a transcript excerpt: Sam Witteveen demonstrates NVIDIA's Dory, a 4B-parameter looped (late…
Based on a transcript excerpt: Sam Witteveen demonstrates NVIDIA's Dory, a 4B-parameter looped (latent recurrent) transformer from NVIDIA's Nemotron Labs that can be downloaded and run locally. The video shows how varying the number of times a layer block loops per token changes output quality and token speed, and notes NVIDIA's own composite score places the model ahead of Gemma 4 26B.
What are the latest AI investment signals?
Latest AI investment signals: 5 funding rounds, 1 market updates, and 2 M&A transactions.
Primary Market – Funding Rounds
| Company | Amount | Round | Investors |
|---|---|---|---|
| GenRobot AI | $100M | Series A+ | SAIC Motor, Eastworlds, Baidu, Shoucheng Holdings, Galaxy General Robotics, GL Ventures, CITIC Goldstone |
| Tutor | $75M | Series B | a16z |
| 42AI | $62.5M | Unknown | |
| Bites | $40M | Series A | PeakSpan |
| Casap | $56M | Series B | Base10 Partners |
Secondary Market – Market Updates
| Ticker |
|---|
| ORCL |
M&A – Mergers & Acquisitions
| Acquirer | Target | Deal Value | Deal Type |
|---|---|---|---|
| Saudi Aramco | Humain | $2.6B | Acquisition |
| Omen AI | Aquatic Labs | $16M | Acquisition |
What are practical AI tips this week?
3 practical AI tips curated from Reddit communities and expert blogs. Willison used an Apollo Global Management chart based on U.S. Census Bureau BTOS survey numbers to d...
Prompt Tips
Willison used an Apollo Global Management chart based on U.S. Census Bureau BTOS survey numbers to demonstrate GPT-5 data analysis and to experiment with Claude's Code Interpreter, then re-ran the numbers to see what changed.
Use published survey datasets (such as the Census Bureau's BTOS) as concrete inputs for testing a model's data-analysis or code-interpreter capabilities, and re-run the same numbers later to compare results.
Productivity
Simon Willison reports occasional confusing Python dependency errors after switching to Python 3.15, where a leftover uv.lock file pins a dependency version that lacks a Python 3.15 wheel (example: watchfiles==1.1.1 failing to build because PyO3's maximum supported version is 3.14).
If you hit a confusing dependency error, delete the uv.lock file and re-run your command (e.g. uv run pytest); Willison says this usually fixes the problem.
Productivity
According to MIT Technology Review, an AI coding agent plans, generates code, runs tests and repeats to correct errors, and a single task can use 10 to 50 times more tokens than a basic chat prompt.
When budgeting or evaluating agentic coding work, account for the multi-step plan/generate/test/retry loop rather than treating it like a single chat prompt, since token use can be far higher.