Are AI models training themselves? Not yet—but Google DeepMind is getting closer. At our AI summit, Google DeepMind chief Koray Kavukcuoglu said the company now trusts AI agents to autonomously run parts of the model-training process: conducting experiments, analyzing results and proposing new hypotheses, all under human supervision. Six months ago, he said, agents weren’t being used this way. #AIAgendaLive #TheInformation @koraykv

Anthropic’s co-founders are seeking majority voting control ahead of an IPO. The Information’s @coryweinberg breaks down the proposed governance changes: "The founders want to have some sort of greater control over their ability to dictate Anthropic’s future. And so they came up with a collective voting structure that gives them 50.1% of the vote over a lot of corporate issues going forward.”

Is there actually a “world model” inside the human brain? Luma AI Co-founder and CEO Amit Jain isn’t convinced—and our own consciousness may make the question even harder to answer. 🚀 Watch the full episode of AI Deep Dive: thein.fo/46yo5px

Tether banking partner EQIBank is fighting to recover about $90 million seized from accounts held by a U.S. payment provider it used to access the American banking system. Read more: thein.fo/4d6e1aV

Exclusive: Anthropic is discussing a deal for up to 1 gigawatt of data center capacity, an investment that could require at least $40 billion. Leasing facilities directly could give it more control over chips and lower its reliance on cloud providers. Read more: thein.fo/4d3OGhH

Exclusive: CME Group’s planned October launch of Nvidia GPU futures is on hold after the CFTC extended its review. Regulators are examining market manipulation risks in a compute market they describe as fragmented and opaque. Read more: thein.fo/4xzu6wZ

Exclusive: GM has started making cathodes for its new lithium manganese rich battery, a key step toward putting the lower-cost chemistry into EVs in 2028. Read more: thein.fo/4hcA5mJ

Google, OpenAI and Anthropic are moving ahead with plans for an AI safety standards body without government oversight. The group could launch by year-end or early 2027 and would set rules for model testing, audits and incident reporting. Read more: thein.fo/4yhkTe6

Open-source models as well as cheaper model releases from Anthropic and OpenAI are helping business customers rein in costs while getting as much or more out of AI, said speakers at The Information’s AI Agenda Live conference in San Francisco Wednesday. Read more: thein.fo/4yg4E0T

DeepSeek’s annualized revenue has reached $1 billion as the Chinese AI startup prepares to raise $7.5 billion at a $75 billion valuation. Read more: thein.fo/3VbqfsB

DensityAI, founded by three former Tesla Dojo leaders, is in advanced talks for funding that would value the year-old AI chip startup at $10 billion. Read more details: thein.fo/4cYWe5u

The Information | TITV | September 25, 2026 nitter.net/i/broadcasts/1wGWjlWWq…

Today on TITV: -DeepMind Gives AI Agents More Autonomy | Koray Kavukcuoglu, SVP, Google DeepMind and Chief AI Architect @koraykv -Google, OpenAI, Anthropic AI Safety Group Takes Shape | @steph_palazzolo & @leomschwartz -Anthropic Co-founders Seek Voting Control | @coryweinberg, Anthropic Reporter -DeepSeek’s Annualized Revenue Hits $1 Billion | @jingyanghk, Asia Bureau Chief 📺 Tune in at 10 am PT / 1 pm ET on thein.fo/46JsPsj

Barrick Gold’s North American unit has signed a five-year AI deal with Avathon ahead of a planned IPO. The contract could generate $10 million to $20 million in annual sales for the startup. Read more: thein.fo/4xJnKeH

Tesla has increased Optimus production roughly tenfold, reaching several hundred robots a week. But problems with the robot’s intricate hands, automated equipment and suppliers are complicating its push to manufacture Optimus at scale. Read more: thein.fo/3TsomHw

Founders Fund and Khosla Ventures have sent partners to China to study AI, robotics and manufacturing companies even as U.S. venture investment there has plunged. The trips underline how much influence Chinese startups now have on American tech strategy. Full story: thein.fo/4hHJvGT

Google is preparing to release Gemini 4 as it works to close the gap with Anthropic and OpenAI. DeepMind chief Koray Kavukcuoglu said the model is in post-training and could arrive well before year-end. Read more: thein.fo/4yfPtor

Exclusive: Anthropic is asking shareholders to approve a new corporate structure that would award CEO Dario Amodei and his six cofounders a special class of shares giving them collective voting control in most corporate matter. Full story: thein.fo/3T8CkOJ

Last scoop of the night: AI chip startup DensityAI is raising hundreds of millions of dollars at a $10b valuation, on the back of a new AWS deal. w/ @_pheebini @validapau theinformation.com/articles/…

TypeSafe AI is talking with investors about raising $1 billion or more as enthusiasm builds around its Jev model and potential valuations above $10 billion. Read more: thein.fo/4hcSXC4