Show only each day's banner story

OpenAI announces that SI already works at the company as a "research intern," while its chief scientist calls on the industry to slow down the same day

SI is already genuinely speeding up the creation of the next SI, the first real step toward self-improving systems; that its own creators say so openly and with numbers means acceleration is no longer a forecast but working reality.

On Sunday OpenAI published two documents. In the "Research acceleration" report the company said it had reached the goal it set last fall, an "automated research intern": agents under human direction carry out clearly defined research tasks that would have taken a specialist several days. For every 8 hours of human work in the research division there are 3.1 workdays of agent work; the median researcher spends more than $600 a day on agents at API prices, and the top 10% more than $7,000. At the same time, more than half of successful 4-8-hour tasks required at least one human intervention; the next goal is an "automated SI researcher" by March 2028. The same day, chief scientist Jakub Pachocki wrote in the essay "An Alien Mind" that no lab has solved alignment and monitoring well enough to keep scaling responsibly at maximum speed for much longer: the reliability of chain-of-thought monitoring is declining, and the current pace could turn into recursive self-improvement. He called for voluntary slowdowns and for turning labs' internal rules into mandatory safety standards with external audits.

Sources 5

openai.comopenai.comunite.aihelpnetsecurity.comen.cryptonomist.ch

Demand for Claude is growing so fast that the company has to buy up energy and chips on the scale of several nuclear plants in advance; compute is becoming the economy's main resource, and bets on SI growth are now measured in hundreds of billions.

Anthropic signs $517 billion of compute contracts in 11 months

Demand for Claude is growing so fast that the company has to buy up energy and chips on the scale of several nuclear plants in advance; compute is becoming the economy's main resource, and bets on SI growth are now measured in hundreds of billions.

According to The Information's tally, since October 2025 Anthropic has signed agreements for at least 14.8 GW of potential capacity worth about $517 billion in total, mostly over the next decade, compared with the $180 billion of spending through 2029 that the company told investors about in December. The largest pieces: Amazon and Google, 11 GW for more than $300 billion; Fluidstack, $50 billion; SpaceX, up to $45 billion; Lambda and Nscale, $80 billion; Microsoft, more than $30 billion for about 1 GW. This is the sum of announced and reported deals, not a single contract; actual payments depend on how much capacity gets built. Anthropic's annual revenue, according to the outlet, exceeds $65 billion, and the company is preparing for an IPO.

Sources 3

theinformation.comdealroom.codevelopmentstoday.com

The word "AGI" has gone from a distant dream to something the biggest players say aloud about a model that has already been released; one can argue about the definition for a long time, but the bar for what counts as "ordinary" SI is shifting right now.

Jensen Huang, after the release of GPT-6 Astra, says "AGI has arrived"

The word "AGI" has gone from a distant dream to something the biggest players say aloud about a model that has already been released; one can argue about the definition for a long time, but the bar for what counts as "ordinary" SI is shifting right now.

The Nvidia chief wrote on X that GPT-6 Astra was trained on roughly 100,000-plus Grace Blackwell NVLink72 systems: "From ChatGPT to o1 and Astra in 4 years. AGI has arrived." He then promised that "400,000 GPUs are on the way." The statement came three days after Astra's release and echoes OpenAI president Greg Brockman's words about the "AGI era"; there is no generally accepted definition of AGI, and OpenAI understands it as autonomous systems that outperform humans at most economically valuable work.

Sources 3

x.comaiweekly.cofoxbusiness.com

The first big money an SI company paid authors for training ran into not technology but the old bureaucracy of the book market; the "SI pays creators" model works, but intermediaries want their cut too.

Publishers and literary agents claim a share of the $1.5 billion Anthropic paid in the books lawsuit

The first big money an SI company paid authors for training ran into not technology but the old bureaucracy of the book market; the "SI pays creators" model works, but intermediaries want their cut too.

Under Anthropic's $1.5 billion settlement, about 500,000 books receive $3,000 each: for books in print the money is split equally between author and publisher, and for self-published books and books whose rights have reverted, 100% goes to the author; the settlement received final approval in July. Now authors complain that publishers are claiming books whose rights reverted long ago (HarperCollins filed a claim for author April Henry on a book whose rights reverted at least 17 years ago), demanding 100% instead of 50%, and literary agents are filing claims without being rights holders. The head of the Authors Guild attributes this to poor record-keeping, while the Writer Beware blogger points to suspiciously identical errors.

Sources 1

techcrunch.com

People who once started the self-driving race are returning to it with new money; competition in robotaxis is growing, which means driverless cars will reach cities faster.

Travis Kalanick's startup Atoms appears to be getting into robotaxis

People who once started the self-driving race are returning to it with new money; competition in robotaxis is growing, which means driverless cars will reach cities faster.

According to the Financial Times, Atoms, founded by Uber's Travis Kalanick, is preparing to hire people and buy companies in autonomous transport and has discussed a robotaxi technology partnership with Uber. This summer Atoms raised $1.7 billion in a round led by Andreessen Horowitz, Uber invested $100 million in it, and Atoms itself bought Pronto, an autonomous mining-equipment startup run by Anthony Levandowski, a former head of Uber's self-driving unit. Kalanick has called this round "unfinished business."

Sources 1

techcrunch.com
Previous day
Follow on Telegram