Wednesday, November 4, 2026 Edition no. 9 · Morning

Artificial intelligence news from Brazil, in English

Noticiário do Joãozinho: Edition no. 9 · Nov. 4, 2026 · Morning

Translated by Joãozinho AI from the original Brazilian Portuguese edition. Facts, numbers and sources are the same as in the original; quotes, video titles and source names stay in their original language. Read the original in Portuguese

Top storyEdition no. 9 · Nov. 4, 2026 · Morning

Google DeepMind researchers push to end lifetime non-disparagement clauses so they can speak about AI risks

The Washington Post reported on Tuesday (11/03) that workers in Google’s AI division — especially at DeepMind in London — are pressing the company to free them from non-disparagement agreements. According to the paper, a revised contract and interviews with two current employees (anonymous) indicate that in the UK the clause can ban criticizing the company for life, even after leaving, and that this chills public warnings about AI risks. Google says it updated terms for new hires, removing the post-termination language, does not apply the clause to former employees, and keeps internal escalation channels. The push gained force after former employee Sofia Liguori told the BBC she ignored the restrictions to speak against military AI.

PDF (pt-BR)

Models

0 stories

No relevant news in this edition.

Research

0 stories

No relevant news in this edition.

Products

2 stories

High importanceNov. 3, 2026

Update: Google confirms — Gemini and Search stay off for K-12 students until the school turns them on; Guided Learning lands in Classroom

In the official Education blog (11/03), Chris Phillips (VP of Education) announced that access to Gemini, Gemini Notebook, Gemini in Classroom and Google Search will be fully off for K-12 student accounts until the school admin explicitly turns them on; details and timeline come in the coming months. In the same post, Google launches Guided Learning in Google Classroom (available globally today): a teacher-assigned activity anchored in class materials, with open questions and steps instead of a final answer, plus real-time insights for the teacher. In Search, Guided Learning enters a pilot in selected US schools later this month. The Washington Post and The Guardian covered the controversy over the Search+AI “bundle.”

Business and Investment

0 stories

No relevant news in this edition.

Regulation

2 stories

High importanceNov. 2, 2026

Apple: dating apps and romantic/sexual virtual companions get a minimum 18+ rating

Apple announced (11/02) new age-rating descriptors in App Store Connect starting 11/16/2026: “Dating and Relationships” (dating/relationship connections between real people and/or a virtual companion with sexual or romantic content) and “Shared Location.” Apps with Dating and Relationships get a minimum 18+; Shared Location may affect the band by region. Required declarations on updates and the questionnaire; all apps by 01/12/2027. In Brazil, local rules ask for A18 for Dating and Relationships and A16 for Shared Location (with a request to MJSP for family cases). TechCrunch and The Verge highlight the impact on Character.AI, Replika and xAI companions.

High importanceNov. 3, 2026

Advocate General Szpunar (CJEU): a RAG chatbot may infringe press publishers’ rights in some cases

In the opinion in case C-250/25 Like Company v Google (Gemini), Advocate General Maciej Szpunar (11/03) recommends declaring the questions on LLM training inadmissible (hypothetical in the case) and, in the alternative, that training an LLM on protected works is, as a rule, allowed by the text and data mining exception. On the case’s focus — chatbot answers — he says there may be infringement when the bot retrieves the publication, reuses it and answers the user: (1) if it accessed unlawfully (e.g. bypassing a paywall/technical measures) and/or (2) if the answer recognizably reuses the form of expression to the point of replacing the original. Bare facts are not protected. The provider is liable if the use is inherent to the business model or if it failed to take effective measures. The opinion is not binding; judgment comes later. Google welcomed the TDM recognition and said Gemini “learns patterns” with controls for publishers.

Open Source

0 stories

No relevant news in this edition.

Agents and Code

1 story

High importanceNov. 3, 2026

Apple warns in its 10-K: AI agents may act in “unintended or unpredictable” ways

In its fiscal 2026 Form 10-K (filed with the SEC on 11/03), Apple wrote that AI agents and other technologies “may take actions autonomously,” including interacting with internal/external systems without authorization, executing or altering code, sending communications and completing transactions; and that they “may behave in unintended or unpredictable ways,” including actions not directed, instructed or authorized. The company says its ability to supervise third-party agents “may be limited.” CNBC ties the warning to the proliferation of agents (citing OpenAI’s Dots and Meta’s Muse) and recent full disk access changes on Mac. On the earnings call, CEO John Ternus said there is “real room” to expand Siri, with a focus on security and privacy.

Local Models

0 stories

No relevant news in this edition.

Finance

0 stories

No relevant news in this edition.

Games

0 stories

No relevant news in this edition.

Brazil

0 stories

No relevant news in this edition.

Social buzz

4 stories

XHigh importanceNov. 3, 2026

Sam Altman: “People should always be in control of the future” — OpenAI against ceding the future to AI

Sam Altman (@sama) wrote in the early hours of 11/04 BRT that “people should always be in control of the future” and that OpenAI opposes relegating humans to “things AI systems think make sense to reward.” The post went viral and pulled the “Ram Altman” thread: after criticism of comments about idle DRAM in PCs, his X display name became “Ram Altman” and he replied “ok yeah” / “i want you to have so much ram.”

Engagement 5.5M views13.8K likes681 reposts1.7K quotes2.9K replies2.6K bookmarks (as of Nov. 4, 2026 08:10)

XMedium importanceNov. 4, 2026

Rumor/tease: Tibo (OpenAI) posts only “ChatBLT” and the timeline starts decoding Bel / Betelgeuse

Thibault Sottiaux (@thsottiaux), an OpenAI engineer tied to Codex/ChatGPT, posted only “ChatBLT.” The timeline treated it as a riddle (BLT = sandwich; Bel/Betelgeuse; “Bye Leaving Tibo”). No official new-model announcement — treat as an engineer tease/shitpost, not a launch.

Engagement 465.5K views3.4K likes109 reposts227 quotes1K replies134 bookmarks (as of Nov. 4, 2026 08:10)

XMedium importanceNov. 4, 2026

Trending on X: Anthropic IPO odds fall on Polymarket after S-1 numbers

An X news story summarizes that prediction markets (Polymarket) reportedly cut odds of Anthropic being the largest IPO of 2026 (from ~62% to ~28%), with SpaceX ahead, in the wake of the S-1/A (9M 2026 revenue ~US$33.24B and compute commitments ~US$895B). Treat as market/X-aggregate echo — the filing figures already ran in prior editions; what’s “new” here is the odds move on social.

XMedium importanceNov. 3, 2026

Trending: “harness engineering” — builders scaffold agents to run for hours

A rising X story describes the shift from loose prompts to “harness engineering”: memory, skills, guardrails, subagents and plugins around the model (kits for Claude Code, Letta, CrewAI). Not a lab announcement — it is builders talking about what keeps an agent from forgetting context or looping.

Videos

2 videos

Teacher's Tech · undefined 2026, undefined

GOOGLE GEMINI'S NEW GUIDED LEARNING - A GAME CHANGER FOR LEARNING!

English demo of Guided Learning in Gemini: the mode guides the student with questions and steps instead of handing over the final answer — the same spirit Google brought to Classroom this Tuesday.

Watch on YouTube

Instruction Partners · undefined 2026, undefined

Gemini in Google Classroom

Overview of Gemini in Google Classroom for educators: useful context to understand the global Guided Learning release in the official 11/03 post.

Watch on YouTube

About this edition

Edition
Edition no. 9 · Morning edition
Date
Wednesday, November 4, 2026
Coverage
Covers what came out since edition #8 (closed around 6:30 p.m. Brasília on 11/03) until around 8 a.m. Brasília on 11/04, including overnight watchdesk leads (WaPo DeepMind, Szpunar/CJEU, Google schools). Closed: 11/04/2026, around 8 a.m. Brasília.
Sources scanned
Official blogs and changelogs (OpenAI, Anthropic, Google DeepMind, Meta, xAI, Microsoft, NVIDIA, Apple, Mistral, DeepSeek, Qwen, Moonshot), arXiv and HF Papers, Hugging Face and GitHub trending, X, press from the US, Europe, China, Japan, India and Brazil, and YouTube.
Criteria
Only news read in a real source, preferably a primary one, gets in, with every link checked. Pages with access blocks (403) were kept; broken links (404) were discarded.