Daily Brief - SiliconNoon (Page 51)

The Arena

Palantir and Nvidia move to restrict Claude over data fears

The AI slowdown fight is drowning out a quieter story: some of the labs' biggest enterprise customers are starting to walk back. And a former president just told Democrats to make AI safety their opening wedge. Palantir, Nvidia and Booz Allen Hamilton are prepared to restrict or drop frontier AI models unless Anthropic and OpenAI guarantee they will not misuse company intellectual property, according to a report that spread across Reuters and The Information. The trigger is Anthropic's June cha

The Arena

Microsoft puts xAI's Grok inside Word, Excel and PowerPoint

A rival model is now living inside Microsoft's own productivity suite, consumer GPUs are being vacuumed up by AI builders, and a former Apple researcher raised $50 million to make AI avatars stop staring at you blankly. Microsoft has added Grok as a model option in Copilot for Word, Excel and PowerPoint. The announcement went up on the Copilot blog on September 12 and widened this week: Grok — sold to enterprise admins under the name SpaceXAI, the corporate identity xAI took after SpaceX absorb

The Guardrails

Trump met Altman privately at the GOP convention, then attacked Amodei

The week's AI slowdown fight turned political — and the two sides are now trading positions, pricing risk, and pulling back their own enterprise customers. Sam Altman asked for the meeting; the President spent Monday attacking the other labs. Trump met privately with OpenAI CEO Sam Altman backstage at the Republican midterm convention on Thursday, according to MS NOW, which cites three people familiar with it; Altman requested it, and the discussion covered AI and its growing power. Two days la

The Everyday

Waymo will run Tokyo's first driverless taxis in 2027

Waymo just committed to a commercial launch date in its second international market, Microsoft quietly let a rival lab's model into Word and Excel, and Oracle started another round of cuts to fund its data-centre build. Waymo, GO and Nihon Kotsu have formally agreed to prepare Japan's first fully driverless commercial taxi service, targeting a public launch in Tokyo during 2027. The three companies announced Tuesday that the service will run vehicles at autonomous Level 4 — no safety driver on

The Arena

Broadcom's CEO says the slowdown call changed nothing about AI demand

The market spent Monday repricing the AI buildout after Dario Amodei's call to pace the frontier. The people who actually write the compute cheques spent the same day saying the number has not moved. Broadcom CEO Hock Tan said the slowdown debate has not touched his AI revenue forecasts. Asked by Jim Cramer on "Mad Money" whether anything in the argument over pacing frontier models had made him reconsider the company's fiscal 2027 and 2028 AI semiconductor targets, Tan answered "No, not in the

The Arena

OpenAI buys Glass Imaging, the AI behind Honor's zoom

Two hardware stories on the same day from opposite directions: OpenAI quietly bought the camera brain it doesn't have, while Apple put a price tag on the camera brain it does. OpenAI has reportedly bought Glass Imaging, the startup whose neural image processing software already ships in Honor smartphones, for more than $300 million. The Wall Street Journal first reported the deal, and neither company has confirmed it. Glass Imaging sells a product called GlassAI — a neural image signal processo

The Frontier

Sakana trains 1000-layer networks with no backpropagation

A paper out of Sakana AI this week takes the oldest objection to local learning and answers it where it bites hardest — depth. Meanwhile the slowdown debate got its most concrete moment yet, on speakerphone. Sakana AI's PC-ALM is the first layer-local training method shown to work on networks up to 1000 layers deep, according to work published by the lab and posted to arXiv. Predictive coding has long been the credible alternative to backpropagation: each layer nudges its own activity to shrink

The Guardrails

Ninth Circuit: the user accessed Amazon, not Perplexity's agent

A federal appeals court has written the first real rule for agents that browse on your behalf, the defense-AI bandwagon is repricing at $20 billion, and the internet's newest spam problem is agents asking permission to exist. The Ninth Circuit vacated Amazon's anti-hacking injunction against Perplexity's Comet browser, holding that the human user — not the company that built the assistant — "accessed" Amazon's computers. The panel's August 4 opinion in Amazon.com Services, LLC v. Perplexity AI,

The Arena

Cornelis raised $205 million. Its pitch: the network, not the chip

Monday's news ran from the inside of the AI rack to the inside of the White House — a networking challenger that says the GPUs are not the bottleneck, a private Altman-Trump meeting that surfaced the same day the two split publicly on slowing down, and a mathematician's attempt to say what the profession should be for. Cornelis Networks announced roughly $205 million in funding alongside Active Compute Fabric, an open networking architecture that puts programmable compute inside the fabric itse

The Guardrails

Carney wants a Financial Stability Board for AI

Monday's governance beat produced something the pacing debate has been missing: an actual institutional design. Then, a workstation card aimed squarely at local inference, and a look inside the cost culture behind China's cheapest humanoids. Canadian Prime Minister Mark Carney called for a global "technology stability board" to oversee AI, modelled explicitly on the Financial Stability Board that coordinates financial regulation. "There's a need for coordination," Carney said in a Bloomberg int

The Arena

Claude gets a wealth-management mode, wired into BlackRock and Schwab

While Washington argued about whether to slow frontier AI down, the labs spent Monday doing the opposite of slowing commercially: packaging agents for industries that had been told their compliance rules made them off-limits. Anthropic released Claude for Financial Advisors, a suite of connectors and workflow skills built around the parts of an advisor's day that aren't client-facing. The plugin bundles skills and connectors into one install and works with BlackRock, Charles Schwab, Addepar, En

The Frontier

Seven PhD students trained a 7B model — and left every log in

Beijing's frontier labs get most of the attention, but the most instructive Chinese AI release this week came from a university project with a student-sized budget and an agent-sized workforce. Seven PhD students at Beijing's Zhongguancun Academy trained a 7B-parameter language model from scratch in one summer, then published everything — code, data recipe, logs, and every intermediate checkpoint. The model is ZGCM-1, a 7.39B dense model released under MIT with a 256K-token context window, aime

See all