Get in touch

A.I. news of Saturday 7 November 2026

Archive From the archive of The Latest A.I. News

This is an archive page with the A.I. news of Saturday 7 November 2026, as it stood that day. The summaries have not been updated since, and a lot may have changed. Today's news is on The Latest A.I. News.

Last updated:

Saturday centers on Anthropic and OpenAI heading toward the market, wobbly AI spending and growing pushback on data centers, with new agents from Meta and Perplexity.

Update: OpenAI reportedly pulls in $11 billion at a $1.25 trillion valuation

High Business

Bloomberg reports that the first $11 billion of OpenAI's new funding round is being wired on Friday, with room for roughly $9 billion more. The backers have not been named. The FT earlier said OpenAI had shrunk the round and trimmed its valuation target from $1.4 trillion to $1.25 trillion.

Update: Anthropic reportedly eyes mid-November listing at a $1,000-1,500 billion valuation

Medium Business

According to CNBC sources (relayed by Bright), Anthropic could list on Nasdaq around mid-November at a valuation of $1,000 to $1,500 billion, below the roughly $2,000 billion discussed earlier. An amended prospectus reportedly lists about $895 billion in obligations, mostly for compute and infrastructure. This is not an official Anthropic statement.

Update: tech stocks rebound after a selloff that pushed the Nasdaq into correction

High Business Chips

After Thursday's heavy AI and chip selloff, which several outlets say pushed the Nasdaq into a correction, tech shares in Asia and the US recovered on Friday. Nasdaq futures rose and chip stocks stabilized. Investors remain wary of data-center spending.

Ramp: corporate AI spending stalled in August and September

High Business Research

Ramp's AI Index shows spending on AI tokens grew only 1.5% across August and September, after a 42% jump in July. Growth picked up again to 8.9% in the month to 23 October. The median company spends just $12 to $15 per employee per month.

Update: GPT-6 Cyber costs more than its predecessor and stays contested; access limited

Certain Models Safety

VentureBeat reports GPT-6 Cyber costs about 60% more for input and 33% more for output than GPT-5.6 Cyber. The model is for now limited to selected organizations and carries OpenAI's internal 'Critical' rating. TechCrunch asks whether that restricted access offers enough safeguards.

Anthropic: research on agent memory, including 'dreaming' where Claude tidies its own notes

Certain Research Models

In an engineering post, Anthropic lays out three layers of agent memory. In 'dreaming', Claude reviews past transcripts offline to correct and organize its memory notes. Anthropic reports Opus 5.5 scoring about 0.84 on CL-Bench with memory versus about 0.66 without.

PSI claims proof of the Aaronson-Ambainis conjecture using a system built on a non-public Anthropic model

Medium Research Models

PSI (Physical Superintelligence) says an internal research system running on a non-public Anthropic model found a proof of a quantum-computing question open since 2008. The proof was checked in the Lean proof assistant. It is the company's own claim and has not been independently reviewed.

US Democrats plan a 'flood' of AI bills from January

Medium Policy

According to Reuters (relayed by Bright), congressional Democrats are preparing a series of bills from January, covering data-center power costs, mandatory independent testing and company liability. These are proposals, not law, and overriding a Trump veto would need a two-thirds majority.

Activists occupy Microsoft's Amsterdam data-center site; police detain 150-200 people

High Policy Chips

Several hundred activists occupied the construction site of a Microsoft data center in Amsterdam's Sloterdijk district on Friday. Dutch police reportedly detained 150 to 200 people on suspicion of trespassing. Microsoft says it does not condone unlawful actions. The protest was part of a two-day 'Stop Dirty Data Centres' campaign across 15 European countries.

AI agents from Google DeepMind reportedly find two weaknesses in legacy 1024-bit RSA

Medium Research Safety

Reports say Google DeepMind AI agents, guided by cryptographers, found two attacks on 1024-bit RSA: signature forgery about 1,600 times more efficient and batch factoring about 30 times cheaper. RSA-2048 is reportedly not affected. Only a secondary source could be checked.

Meta adds 'Spaces' to its Muse agent for project chats that keep updating in the background

High Products

Meta is launching Spaces in Muse: users create project-specific conversations that can connect to third-party apps and update in the background. According to TechCrunch, Meta cited roughly 10 million Muse users in the US and Canada in October.

More than 800 AI-driven Artex hacking tools found exposed on the web without a password

High Safety

Censys researchers found more than 800 Artex deployments reachable from the web without a password. Artex is an open-source, China-made tool that probes for vulnerabilities autonomously and has been linked to attacks including one on the Korean finance industry and one on ConnectWise. Its makers have said they will close the source.

Anthropic tests 'Personal Finance' for Claude: analysing bank and card transactions (US, beta)

High Products

According to Bright, citing Anthropic, a 'Personal Finance' beta is starting for Claude Pro, Max and Ultra subscribers in the US. Claude can read up to two years of linked bank and card transactions, read-only, and Anthropic says the data is not used for training. Anthropic warns of errors and says it is not investment advice.

Crusoe sells unfinished Wyoming data-center project to Google, which plans a $17 billion investment

High Chips Business

Crusoe confirmed it sold an unfinished data-center project in southeastern Wyoming to Google, which plans to invest $17 billion there. The price was not disclosed. The project was originally planned at 10 gigawatts, and Bloomberg first reported the sale in June.

Perplexity launches 'Teammates': always-on AI agents for $200-a-month Max subscribers

Medium Products

Perplexity is introducing 'Teammates', persistent AI agents that work on tasks around the clock. They are aimed at subscribers of the $200-a-month Max plan. Details could only be checked through secondary reports.

Nvidia touts its Vera CPU for top throughput and fastest agents; Box builds on Nemotron

Certain Chips Products

Nvidia's blog says its Vera CPU delivers the highest throughput and fastest AI agents; these are Nvidia's own claims. Box is using Nvidia's Nemotron models for a new agent-first system for retrieving business content.

Update: Mistral Studio to get private cloud and on-premises options in late November

Certain Products

After launching Mistral Studio for enterprise agents, workflows and apps, Mistral says private-cloud and on-premises versions follow in late November. That fits its stress on control over models, data and infrastructure.

Alibaba's Qwen posts a cryptic teaser for Qwen4, with no name, date or numbers

Low Models

Alibaba's Qwen team posted a cryptic teaser for a possible Qwen4 model. No name, date or benchmark figures were given, so it remains speculation for now.

Agility becomes the first pure-play humanoid robot company on the stock market

Medium Business Other

TechCrunch reports that Agility is positioned as the first publicly listed company focused solely on humanoid robots, and asks whether investors will buy in. It fits the wider wave of physical AI.

Nothing to release its first AI gadget next month

Medium Products

Bright reports that Nothing's first AI device arrives as early as next month. What exactly it can do is still largely under wraps.

Apple Machine Learning: faster retrieval for more efficient RAG

Certain Research

Apple's research blog describes a method to find the best-matching vector among millions of candidates faster (maximum inner product search), the basis of search, recommendation and RAG systems.

Google DeepMind: agentic cheminformatics with Google Antigravity

Certain Research Products

Google DeepMind has published a blog on using AI agents in Google Antigravity for cheminformatics, the computational analysis of molecules.

USA Today blames AI search for part of its traffic and digital revenue decline

Medium Business Other

Poynter reports that USA Today Co. attributes part of its third-quarter drop in web traffic and digital ad revenue to AI search features. Publishers are increasingly feeling the impact of answer engines.

Next up: Around mid-November: a possible Anthropic Nasdaq listing (per sources)

All A.I. news

The short version

Saturday 7 November 2026

  • Anthropic reportedly eyes a mid-November listing at a $1,000-1,500 billion valuation, below the roughly $2,000 billion discussed earlier.
  • Pushback on data centers grows: activists occupied a Microsoft site in Amsterdam and US Democrats plan a wave of AI bills from January.
  • Anthropic published research on agent memory, and PSI claims a proof built on a non-public Anthropic model (not yet independently reviewed).
  • Meta adds Spaces to its Muse agent and Perplexity launches 'Teammates'; over 800 Artex hacking tools were found exposed online.

What does TL;DR mean?

TL;DR stands for “Too Long; Didn’t Read”. It’s the name for a short summary for people who are short on time. Click the button for the key points of this day in a few lines; all of the day’s items are right below.

Archive

Every day has its own page. They stay up for three months.

November 2026

October 2026

September 2026

Everything here is a summary of reports from other media, not original research. The label (Certain, High, Medium, Low) is my estimate of how reliable the source is. The summaries are generated automatically with AI and may contain mistakes; read the article itself for the details (the star marks the leading source). Sources: leading A.I. and tech sites and the official blogs of the A.I. companies.