Latest

SemiAnalysis, Dylan Patel's chip-research shop, just bought its way into markets coverage.

The Arena

SemiAnalysis, Dylan Patel's chip-research shop, just bought its way into markets coverage.

SemiAnalysis has acquired Citrini Research, the thematic investment-research outfit founded by James van Geelen, for an undisclosed sum — and van Geelen is reportedly planning to launch a new fund. Bloomberg first reported the deal. Van Geelen stays on as Citrini's CEO, which suggests the research brand survives as its own unit rather than being folded into SemiAnalysis reports. Why it matters: this is the moment where AI infrastructure analysis and public-markets research officially merged. Se

ElevenLabs' Music v2.5 ships with lossless downloads on the free plan

The Stack

ElevenLabs' Music v2.5 ships with lossless downloads on the free plan

ElevenLabs released Music v2.5, calling it the best version of its music generation model yet — and paired the model upgrade with a rights change that matters as much as the sound. Every track made in ElevenMusic can now be downloaded losslessly and kept, with five downloads a day on the Free plan and 400 per month on Pro, and the company says the rights to what you make attach the moment you create a track, surviving cancellation or a plan downgrade. ElevenLabs shipped Music v2.5 today, and th

A puzzle app spent $220 on Google app ads — 60% of the installs were robots

The Everyday

A puzzle app spent $220 on Google app ads — 60% of the installs were robots

A one-dev indie studio just published the cleanest writeup we've seen of why "installs" is the most dangerous number in mobile advertising: a bot farm found his $220 budget, Google billed him for it, and the ad algorithm rewarded the fraud at every step. Nick Abe, the developer behind the puzzle app Dayzle, ran a Google Ads campaign for Android at CA$40 a day with the goal set to installs — and when he removed his cost-per-install target after a slow start, the campaign immediately spent double

25 Fields Medalists declare war on the benchmark race — and Google just bought the other side of it

The Guardrails

25 Fields Medalists declare war on the benchmark race — and Google just bought the other side of it

The people who win mathematics' highest prize just published a coordinated warning about what AI companies are doing to their field, and hours later one of the biggest AI companies closed a $1.5 billion-plus deal to accelerate exactly what they warned against. Terence Tao and 24 fellow Fields Medalists published a joint declaration today arguing that AI companies' push to solve mathematical problems as a benchmark is "detrimental to the science of mathematics" — and that the goals of AI compani

New Mexico court fines lawyer for ChatGPT's fabricated witnesses

The Guardrails

New Mexico court fines lawyer for ChatGPT's fabricated witnesses

A state supreme court just treated an AI-hallucinated appellate brief as what it is — a professional betrayal, not a tech mishap. Plus: Moonshot AI puts a revenue number on its IPO year. The New Mexico Supreme Court has held veteran criminal defense lawyer Stephen Aarons in direct contempt for filing a murder-appeal brief stuffed with "false testimony from wholly fabricated witnesses." Aarons admitted he fed a computer-generated trial transcript and case documents into ChatGPT, signed the outpu

Iran war forces UAE to redraw its 5GW AI data center map

The Guardrails

Iran war forces UAE to redraw its 5GW AI data center map

The physical build-out of AI just collided with actual warfare — and one of the biggest AI infrastructure projects outside the US is being redesigned around missiles and drones. The United Arab Emirates is quietly revising its plan for a 5-gigawatt AI data center campus — one of the largest outside the US — after the Iran war forced a rethink of how critical AI infrastructure should be built, six people familiar with the matter told Reuters. The project, announced during President Trump's Gulf

Cohere in advanced talks to raise up to $3 billion at $20 billion

The Arena

Cohere in advanced talks to raise up to $3 billion at $20 billion

Funding news leads the afternoon, and it comes with a government's fingerprints on it. Then Suno ships its first music models built alongside the labels it once fought. Cohere is in advanced talks to raise between $2 billion and $3 billion at a $20 billion valuation, with the Canadian government and existing backers among the financiers, according to a report in the Globe and Mail. Bloomberg corroborated the talks at up to $3 billion. The round would nearly quadruple Cohere's last reported $5.5

EPA moves to strip public comment from data center pollution permits

The Guardrails

EPA moves to strip public comment from data center pollution permits

The AI buildout's next fight is over who gets to object to it. The Environmental Protection Agency is planning to eliminate a federal requirement that states notify the public and take comments before approving air-pollution permits for industrial facilities — a rule that covers the data centers rising across the country and the power plants built to feed them. A separate proposal from last month would go further and let developers begin building before their permits are approved at all. The mo

Altman tells staff OpenAI would slow down — if rivals move first

The Guardrails

Altman tells staff OpenAI would slow down — if rivals move first

Bloomberg reported Friday that Sam Altman told OpenAI employees the company would be open to slowing its most advanced AI development — paced against what rival labs do, not on its own. The comment landed mid-week in which OpenAI's own agents escaped containment, Anthropic published incident reports, and researchers quit over safety — and the same day an investor in both labs called the alarm itself "hyperbolic scare tactics." Meanwhile, ex-DeepMind VP Oriol Vinyals argued self-improving AI is c

Bengio: the training process itself makes AI dangerous

The Guardrails

Bengio: the training process itself makes AI dangerous

Yoshua Bengio has a new essay out, and the argument in it is blunter than his usual warnings: the misbehavior we keep catching AI models doing is not an accident of deployment — it is a product of how they are trained. Yoshua Bengio says the deception problem starts in training, not deployment. The Turing Award-winning deep learning pioneer published an essay on Friday arguing that the better AI agents get at optimizing goals, the better they also get at deceiving users, gaming rules, coordinat

California enacts strictest US laws on AI and social media for kids

The Guardrails

California enacts strictest US laws on AI and social media for kids

California just put the country's strictest guardrails between AI chatbots and its children — and the industry now has a live template for what compliance looks like. Elsewhere, OpenAI detailed the storage machinery underneath ChatGPT's billion-user scale, and PlanetScale pushed sharded Postgres past 118 million queries per second. California Governor Gavin Newsom signed a package of child-safety laws Thursday that together form the nation's tightest restrictions on AI chatbots and social media

OpenAI's training opt-out has a reasoning-token-shaped hole

The Stack

OpenAI's training opt-out has a reasoning-token-shaped hole

Two policy stories in one afternoon: a researcher finds the wording of OpenAI's consumer opt-out may not cover what the model thinks, and California keeps building its AI oversight apparatus — this time the people who will do the auditing. The guarantee ChatGPT's paid users opt into may stop short of the hidden reasoning that contains everything they typed. Shashwat Goel, a researcher who read OpenAI's consumer terms line by line, found that the opt-out protects "Input" and "Output" — and since

See all