Jensen Huang reportedly weighs a return as Nvidia CEO, Anthropic IPO orders reportedly reach six times the shares on offer, and more - Daily AI Brief #584
In this edition, covering 14 to 16 November:
- Jensen Huang has discussed returning as Nvidia CEO, the Financial Times reports, while Treasury Secretary Scott Bessent says the board should pick an outside CEO.
- Institutional investors ordered about six times the available shares by the fourth day of Anthropic's US IPO roadshow.
- An internal Amazon document describes World Product Graph, a project to collect data on nearly every product for sale, Reuters reports.
From our lists
An AI Brief list covers companies that audit and certify AI systems.
Nvidia
Jensen Huang, who resigned as Nvidia CEO on 12 November, has discussed returning to the job with confidants, the Financial Times reports. One Huang ally told the FT they are putting together a war room of Silicon Valley executives to lobby for his return and would seek support from the White House. Huang comeback talks
Altimeter Capital CEO Brad Gerstner posted a letter, also signed by Atreides' Gavin Baker, that urges the Nvidia board to pursue a slow, measured and deliberate leadership process that includes the potential return of Huang. Altimeter and Atreides letter
A public petition asking the Nvidia board to bring back Jensen Huang during an independent investigation passed 115,000 signatures. Signers include about 6,000 current employees, 63,000 shareholders and Elon Musk, TechCrunch reports. Petition for Huang
Treasury Secretary Scott Bessent called for a dramatic shakeup at Nvidia and said the board should look outside the company for its next CEO. He also questioned whether Nvidia pursued illicit sales of its advanced AI chips to boost earnings. Bessent
Commerce Secretary Howard Lutnick favours narrower limits on overseas access to advanced US AI chips, The New York Times reports, while his former firm Cantor Fitzgerald works with data center companies such as Bitdeer. NYT on Lutnick
Governments in China, Singapore, South Korea, Taiwan and Japan responded to Huang's departure on chip supplies, AI projects, planned facilities and investigations linked to former Megaspeed executive Alice Huang. South Korea said Nvidia reaffirmed its plan to supply 260,000 advanced GPUs, and Taiwan said Nvidia's planned Taipei headquarters remains on track. Asian governments respond
Malaysia is reviewing legislation that would require data center operators to report imported GPUs, power use, facility locations and customers, and to pay for reserved electricity capacity even when it goes unused. The measures are meant to stop AI chips from being rerouted to countries such as China, although power data alone may not reliably show how many GPUs a facility holds. Malaysia GPU reporting law
Anthropic
Institutional investors ordered about six times the available shares by the fourth day of Anthropic's US IPO roadshow, after earlier uncertainty over whether the roadshow would go ahead. Investors still question whether Anthropic can keep its lead over Google and other AI companies, broaden Claude beyond coding and meet revenue forecasts that differ widely. IPO demand
Anthropic hired 33 counterintelligence staff in 2026, up from 14 in 2025, according to Revelio data compiled by Reuters. OpenAI hired 22, up from nine. Counterintelligence hiring
Amazon
An internal Amazon document describes World Product Graph, a project to collect data on nearly every product for sale, with Coca-Cola and Bayer contributing, Reuters reports. Reuters describes the plan as a way to counter AI shopping agents. World Product Graph
Apple
Apple launched the Apple Health AI Usage Study with OpenEvidence, a virtual study and app that lets up to 200,000 US participants ask health questions and examines how far they trust and understand the answers. Health AI study
Apple plans to announce its first smart glasses, without augmented reality and built around a revamped Siri, at WWDC next June, Bloomberg's Mark Gurman reports. Smart glasses
Baidu
Baidu's AI Cloud Infra revenue rose 94% to RMB 8.2 billion in the third quarter, while its total revenue fell 1% to RMB 30.8 billion. Baidu results
Models and research
OpenAI says GPT-6 Astra led the detailed proof behind a candidate lattice definition of the Standard Model of particle physics. Physicist Nathaniel Craig credits the model with key construction and proof steps; practical simulations and important consistency tests remain open. GPT-6 Astra proof
Agnes AI released Agnes 2.5 Pro Theta, a 188B parameter multimodal mixture of experts model with open weights under the Apache 2.0 license. The company reports 92.9 on GPQA Diamond and 90.5 on LiveCodeBench v6. Agnes 2.5 Pro Theta
Alibaba launched Qwen4-Max, a closed 2.4 trillion parameter flagship model. Alibaba says it matches or beats OpenAI's GPT-6.1 Sol and Anthropic's Opus 5.5 on coding tests and cuts average task costs on its own benchmark by about 60%. Qwen4-Max
Domyn, which leads the EUROPA consortium, trained the EUROPA model at near one-trillion-parameter scale on Germany's JUPITER supercomputer. Domyn says it raised training throughput 14-fold on publicly funded compute and has published its methods. Domyn EUROPA
UK Research and Innovation named 20 new Turing AI Fellows, backed by more than £39 million, for core AI research and uses from disease diagnosis to clean energy. Turing AI Fellows
Data centers
Telus will no longer develop two sovereign AI data centres in Vancouver with Westbank. Westbank says both projects continue with a new partner it did not name. Telus and Westbank
A Scottish Greens campaign recorded 13,146 objections to CX Tech Limited's proposed 1,000MW AI data centre campus at Chapelcross in Dumfries and Galloway, which would be the world's largest data centre campus if approved. Campaigners raised concerns about energy use, building on farmland and the environment; CX Tech says the project would use dedicated renewable energy connections and create jobs. Chapelcross objections
Samsung SDS is in early talks to sign Anthropic and OpenAI as tenants for its Korean AI data centres, the Financial Times reports; none of the companies has confirmed. Samsung SDS talks
Policy
The Trump administration plans to name more than two dozen US AI companies, among them Nvidia, OpenAI and Anthropic, as national champions to help them sell abroad and compete with Chinese rivals, The Wall Street Journal reports. The plan has not been confirmed. National champions
Japan, Australia, Israel, the UK, South Korea, India, Qatar and the UAE signed the Pax Silica San Francisco Declaration with the United States and agreed to draw up a joint list of critical AI technologies and supply chains that need more protection from China's influence, the Financial Times reports. Pax Silica declaration
The United States, Russia and Israel are expected to oppose a push at a UN weapons convention meeting in Geneva this week to start negotiations on legally binding rules for lethal autonomous weapons, Reuters reports. Autonomous weapons talks
Security
Cursor says an access key exposed by Fireworks AI let an attacker copy coding agent traces containing credentials, reach Cursor's SaaS tools and plant the malicious Claude Code extension that Cursor removed earlier. Cursor breach
Cohere completed its acquisition of AI security startup AISLE and launched Cohere Cyber with a hub in Prague. It also bought AI code review startup cubic to build a secure coding unit. Cohere Cyber
An international investigation led by Japan's JPCERT/CC found about 1,000 servers, most of them hosting AI tools such as ARTEX, that have been online since October and were apparently used to search for vulnerabilities and attack websites worldwide. JPCERT/CC servers
Surveys
In the Federal Reserve's November Financial Stability Report, the 23 market contacts surveyed cited AI most often as a risk to US financial stability, pointing to stock valuations, AI enabled cyberattacks and AI financing. Fed survey
A Lowy Institute poll of 3,028 Australians found that 65% think the risks of AI outweigh its benefits, and only 23% are comfortable with businesses using Chinese AI. Lowy poll
In Forrester's State of AI Survey 2026, 82% of respondents report at least one generative AI application in production, but only 20% of AI decision makers say their organization has the technology to scale its AI initiatives. Forrester survey
Courts and police
Maricopa County Judge Daniel Fenzel dismissed Anthony Quidera's DUI case after an Arizona Department of Public Safety trooper disclosed using Truleo AI to help write the probable cause statement. Quidera was released and did not attend his scheduled arraignment. Arizona DUI case
Delaware Vice Chancellor Lori W. Will sanctioned Richards, Layton & Finger partner Richard Rollo after a court brief contained fabricated quotations, inaccurate citations and unsupported legal claims generated with an AI tool. Rollo must pay the $2,500 penalty personally; the court declined to sanction his law firm. Delaware sanction
The Royal Canadian Mounted Police in Prince Edward Island say AI generated child sexual abuse material appears in about one third of their online child abuse investigations. Police in St. John's, Newfoundland and Labrador, charged a 17 year old in a case involving AI altered images of minors. P.E.I. RCMP
The top scorer on South Korea's tax accountant exam admitted that he photographed questions with a hidden camera, sent them to an AI system and read the answers on a phone. Seoul police are investigating. Tax exam cheating
Ethnic minorities made up 24% of the people South Wales Police checked with officer initiated facial recognition in 2025, against 7% of the area's population, according to figures released under Freedom of Information. South Wales facial scans
Health
The NHS is trialling Ufonia's Dora Care, a generative AI telephone assistant that monitors patients with osteoporosis and supports clinicians. The voluntary two year project involves Newcastle University and the University of York, with safety monitoring and funding from Innovate UK. The calls supplement existing care, and the system could later support patients with diabetes, heart failure and COPD. Dora Care trial
Wolters Kluwer Health agreed to acquire San Francisco based AgileMD, which makes AI clinical decision support software, for up to $100 million, with $40 million upfront and the rest tied to performance targets. AgileMD's 23 employees will join Wolters Kluwer Health, and the deal is expected to close in the first quarter of 2027 or sooner, subject to approvals. Wolters Kluwer and AgileMD
Also
Google's Gemini 4 Carbon ranked first on the MathArena maths benchmark with 85.4% plus or minus 3.0%, ahead of Anthropic's Claude Fable 5.2 and OpenAI's GPT 6.1 Sol. The score ranges overlap, so the ranking is not conclusive, and the tested run cost substantially more than the other models at their selected reasoning settings. MathArena
Former OpenAI board member Helen Toner told The Verge she would vote again to remove Sam Altman, as the board did in 2023, and that AI safety cannot rest on one person. Helen Toner
Thomson Reuters added regulatory intelligence from CUBE, covering more than 150 countries, to CoCounsel Legal through a link between the two companies' AI agents. CUBE and CoCounsel
Insurance law firm Wotton Kearney is rolling out Harvey across its 13 offices in Australia, New Zealand, Singapore and Thailand after a 200 user pilot reached 94% adoption. Wotton Kearney and Harvey
Weil Gotshal incubated Kepler, an AI native law firm that plans to raise hundreds of millions of dollars, do M&A work on fixed fees and launch in early 2027, PitchBook reports. Kepler
Tesla is offering select customers in its Robotaxi cities a Cybercab for personal use at $29,990 plus a mandatory $299 monthly Autonomy Subscription, without which the car cannot drive. Tesla Cybercab
Pony.ai and Toyota put the bZ4X robotaxi, a Gen-7 model they developed together, into fully driverless paid service in Guangzhou; the partners are working toward about 1,000 bZ4X robotaxis in China. Pony.ai and Toyota
The California Department of Insurance started a survey of insurers on their AI use and scheduled a public workshop for 15 December to inform future consumer safeguards; it has not set rules. California insurance survey
San Francisco based military robot startup Foundation has not paid employees since it missed its 30 October payroll, CNBC reports. CEO Sankaet Pathak, the company's sole board member, borrowed $4.6 million from Foundation in 2025 to buy a home; its reported $24 million in Pentagon contracts came with its acquisition of Boardwalk Robotics. Foundation payroll
ICEYE unveiled its fifth generation of radar satellites, which add an AI processor that screens data in orbit and offer resolution as fine as 11 cm for government customers. ICEYE Gen5
Climate intelligence startup mistEO raised Rs 18 crore in a seed round led by MGF-Kavachh to expand its AI weather and climate risk platform across Asia. The Manchester Animation Festival in the UK will not accept films made with generative AI from 2027. mistEO · Manchester Animation Festival
Indian startup Bharat Semi raised a $24 million seed round led by Info Edge, South Park Commons and Indus Bridge Ventures for its AI image sensors, and says it has design wins with General Atomics and Kaynes Semicon. Bharat Semi
In European funding, Intergrid raised €23.5 million in a round led by Kvanted and Tesi for Autopilot, which runs energy systems at more than 100 industrial sites, and Daisytuner raised a €6 million seed round led by Elaia and LEA Ventures for software that makes AI models run faster on different chips, and launched its Daisyflow inference engine. Valeria raised a €5.5 million seed round led by BlackFin for an AI payroll platform that links to Spanish public authorities, Stoa raised €3.5 million in pre seed funding led by Lifeline Ventures for AI software used on more than 25 Nordic construction and data center sites, and Treat raised €3 million in pre seed funding led by byFounders and Illusian for AI intake that briefs doctors before digital consultations. Intergrid · Daisytuner · Valeria · Stoa · Treat
Flagship Pioneering committed £100 million to a London lab, its first company creation unit outside the US, which will build AI native platform companies and hire about 50 people in the UK. Flagship London lab
We hope you enjoyed this article
Consider subscribing to one of our newsletters like Daily AI Brief.
Also, consider following us on social media:
Daily AI Brief
Daily report covering major AI developments and industry news, with both top stories and complete market updates