AI Morning Brief · Nov 2, 2026

Coverage window: 2026-11-01 07:00 → 2026-11-02 07:00 (Beijing Time)

Research note: English edition written primarily from English-language sources; some items cite Chinese-language sources.

Top Stories

1.Four labs back one robots.txt AI-training opt-out

a shared `__usage_ai-training` policy token announced Oct 30 by OpenAI, Google, Microsoft and Anthropic, from an IETF personal draft by Anthropic's Kevin Kelley; Anthropic has started work, Microsoft starts Feb 1, 2027, Google plans early 2027, OpenAI says "in the coming months", Meta has not committed (Nov 1)

2.Judge lets NAACP suit over xAI's gas turbines proceed

a Mississippi federal judge ruled Oct 30 that trailer-mounted turbines at xAI's Southaven Colossus 2 site are stationary sources, not exempt mobile sources, rejecting xAI's bid to dismiss; at least 35 turbines are alleged to have operated without permits, and an injunction request is still pending (Nov 1)

3.DeepSeek open-sources a super-agent prototype

dsh-superagent-prototype, released Oct 31 on GitHub under the MIT license, stays online, watches GitHub and Feishu team chats and delivers work as pull requests; DeepSeek's own synthetic test reports a 12.4-minute median vs 13.8–20.4 minutes for a single session with a subagent tool, but about 23.2 million cached tokens per run vs 6.1–10.8 million (Nov 1)

4.Anthropic IPO timetable updated: roadshow in the week of Nov 9, trading before Thanksgiving

Bloomberg reporting relayed Nov 1; investor discussions put valuation around $1.8–$2 trillion, still unconfirmed, and the S-1 pricing range and share count remain blank with no amended filing yet (Nov 1)

5.Samsung executives met OpenAI's Altman on Oct 30, no deal announced

Bloomberg reporting relayed Nov 1 says Altman met about 50 Samsung executives at Samsung's Yongin center to discuss AI strategy; no cooperation, order or investment was announced (Nov 1)

6.Japan GPU prices jumped 18%–45% in a month

a Japanese monthly retail survey dated Nov 1 puts the RTX 5090 up 45% to about $8,100, the RTX 5080 up 32% and the Radeon RX 9070 XT up 25%, with depleting inventory blamed and a late-September open-source project, Strata, cited by market participants as a trigger for on-premise AI buying (Nov 1)

7.AI tokens go on sale for Singles' Day in China

Tmall's token category, live since Sep 3 with DeepSeek, Alibaba Cloud, Zhipu and Kimi onboard and MiniMax opening the first brand flagship store, saw sales up 450% month-on-month on the Oct 20 opening day; a Qwen AI Lite monthly card was listed at 39 yuan vs 60 yuan official (Nov 1, Chinese-language source)

8.Two security disclosures: Chainalysis breach and VOLTA malware's three-model chain

Chainalysis says an attacker used "offensive agentic techniques" to reach internal systems, accessing names, emails, user IDs and role labels, with no evidence so far of exposed investigation content, credentials or session tokens; Cisco Talos reported Oct 31 that VOLTA infostealer sends stolen data to a model chain of Claude Opus 4.6, Kimi K2 and Gemini 2.5 Flash to rank victims (Nov 1)

Platform & Model Tracking

OpenAI (GPT/ChatGPT)

Shared robots.txt training opt-out (see Top Story 1), support promised "in the coming months", with no stated rule for conflicts with GPTBot rules. Samsung meeting (Top Story 5) produced no announced deal. No new GPT model, pricing or ChatGPT availability announcement in the window.

Anthropic (Claude)

IPO timetable updated (see Top Story 4): roadshow reportedly in the week of Nov 9, trading before Thanksgiving on Nov 26. The IETF robots.txt draft was submitted by Anthropic's Kevin Kelley and Anthropic has begun development. No new Claude model, pricing or availability announcement in the window; an answer to Senator Warren is due Nov 3.

Google DeepMind (Gemini)

Shared robots.txt signal planned for early 2027, to be treated as disallowing Google-Extended for future Gemini training and grounding. No new Gemini model, pricing or availability announcement in the window; Gemini 4 Carbon general availability date, API model ID and intro-period length remain unannounced.

Meta

On the shared robots.txt signal, Meta only said it supports publishers expressing preferences and did not commit to honoring the signal. No new Llama or Muse model, pricing or availability release in the window.

xAI (Grok)

NAACP turbine lawsuit proceeds after the dismissal bid was rejected (see Top Story 2). No Grok 5 release and no pricing change in the window; the Grok API watermark default takes effect Nov 7 as previously announced.

Microsoft (Copilot/Azure AI)

Shared robots.txt signal from Feb 1, 2027, with NoArchive and NoCache to stop counting as training signals after Mar 1, 2027, plus a separate page-level AI Grounding preference. Its US school AI safety and privacy framework reached its scheduled application date Nov 1. A Washington Post test relayed Nov 1 found Copilot cited the AI-produced Ballot Brief in 75% of voter-guide tests and ChatGPT in 22%.

NVIDIA

No new chip release in the window. The Japan GPU price surge (see Top Story 6) is a retail-market fact, not an NVIDIA price announcement. The reported Vera Rubin bandwidth revision (HBM4 22 to 19.2 TB/s) is a September product-page edit relayed Nov 1, not a new November announcement.

Mistral

No new model release and no pricing change in the window; Mistral has not committed to the shared robots.txt signal.

China AI Companies

DeepSeek

Super-agent prototype open-sourced Oct 31 (see Top Story 3), a research prototype, not a product launch. No new model or funding announcement in the window. Token sales on Tmall (see Top Story 7) are a distribution fact.

Alibaba (Qwen)

The Apple Intelligence in China timetable was restated Nov 1 — developer beta Nov 4, public release with iOS 27.3 in December — with no new announcement. Qwen AI Lite monthly card listed at 39 yuan in Singles' Day promotion (see Top Story 7). Qwen4 remains unannounced.

Moonshot (Kimi)

Kimi K2 appears in the VOLTA malware fallback chain reported by Cisco Talos (see Top Story 8), a research finding, not vendor cooperation. Kimi K3.1 Pro weights remain unpublished ahead of the Nov 4 deadline. Kimi is onboard Tmall token sales (see Top Story 7).

Zhipu (GLM)

Onboard Tmall token sales (see Top Story 7). No new model or pricing release in the window.

ByteDance (Doubao)

No new in-house model release in the window; no new model added to Doubao Work's third-party list.

Tencent (Hunyuan)

No new model or product release in the window.

Baidu (Ernie)

No new product fact in the window.

MiniMax

Opened the first brand flagship store for token sales on Tmall (see Top Story 7). No new model release in the window.

Manus

No material new development in the window.

Worth Reading

1."Google, Microsoft and OpenAI say they will honor one AI training opt-out" — Pivot News, 2026-11-01

the full robots.txt picture — IETF personal-draft origin, four different timetables, Meta's non-commitment and the future-training-only limit.

2."AI News Nov 1, 2026: xAI Turbine Lawsuit, DeepSeek Agent" — The Diary of AI, 2026-11-01

a quiet Sunday's two main threads — the xAI turbine ruling and DeepSeek's super-agent prototype — plus the Nov 3 and Nov 4 follow-up deadlines.

3."AI Daily Brief: 1 November 2026" — Precise Impact, 2026-11-01

the Chainalysis breach, the VOLTA three-model chain, the Qwen–Apple Nov 4 date, the xAI suit and Microsoft's school framework in one cross-check.

Unverified Items

1.

Google Gemini "hacking three companies then stopping" in an Irregular test: only an Al Jazeera mirror site relay, citing a Wall Street Journal report, with no WSJ original or Google statement verified — single source, unverified.

2.

GPT-6 Astra Law API pricing ($10 input / $50 output, promotional through Dec 31): only TestingCatalog / Observatory relay, no OpenAI pricing-page original verified — single source, unverified.

3.

NVIDIA Vera Rubin bandwidth cut (HBM4 22 to 19.2 TB/s, NVLink 6 3.6 to 3 TB/s): a September product-page edit relayed Nov 1, with no NVIDIA statement verified — unverified as a November announcement.

4.

Meta humanoid robot prototype with China's Fourier: only an Engadget relay via The Diary of AI sidebar, no Meta announcement verified — single source, unverified.

5.

Google AI Mode hurting Search monetization: only a Latent Space relay of a Google business chief's remark, no earnings-call original verified — single source, unverified.

6.

Outstanding items with no new progress in the window: Grok 5, Qwen4, Anthropic S-1 pricing range and share count, Kimi K3.1 Pro weights due by Nov 4, Zhipu "3.5 trillion parameters" rumor, RTX PRO 5500 orders, Ant Ling/Ming single-source reports, ten-year memory contracts, Gemini 4 Carbon general availability, NVIDIA Nemotron 3.5 Transcribe official blog, DeepSeek/Moonshot distillation allegation, Meta parental-reminder extension, xAI Mississippi home-buying report, NVIDIA–Nscale guarantee analysis, Grok Imagine Video 1.5 Lite, OpenAI first-close amount, Anthropic's answer to Senator Warren due Nov 3, Apple iOS 27.3 developer beta due Nov 4. The shared robots.txt signal, listed as unverified in the previous issue, is now confirmed and removed from this list.