Good morning. What’s the deal with AI spending—will it keep going up up up, or come down from those highest of highs in the years ahead?
According to the latest from Brookings, the spending is still massive, even if its rate of growth is slowing. The think tank projects U.S. AI spending on data centers and related AI infrastructure to total $10.3 trillion between last year and 2032, or an annual average of 3.6% of GDP.
That’s apparently a greater share of American GDP than the initiatives to build canals, railroads, and the power grid combined. From an investment standpoint, the AI buildout is bigger than the one that gave us the interstate highway system; it’s also bigger than what gave rise to the consumer Internet.
Not everything about that is good. It’s a lot of economic eggs in one basket, for starters. It’s a lot of inflation, too. And an increasing share of the AI boom is being financed by debt…which will be a real problem in the event of a downturn.
The news follows. Have a wonderful weekend. —Andrew Nusca
Thoughts? Suggestions? Hot tip? Drop me a line.
THE BIG STORY
OpenAI’s rogue AI agent issue is bigger than it seems
OpenAI’s issues with rogue AI agents are more extensive than the company has previously acknowledged—and may be ongoing. That is the conclusion of a new report from an independent research firm.
The new revelations emerged on the same day the Australian government said OpenAI’s rogue AI agents had hacked an agency that held the country’s Medicare data.
The latest findings come from Transluce, an independent non-profit research lab focused on AI oversight. It said Wednesday that it discovered OpenAI agents attacking additional Australian government websites, including its Institute of Health and Welfare, as well as BOSCAR, the crime statistics body for the Australian state of New South Wales.
In addition, it discovered at least two previously unreported incidents of OpenAI’s agents attacking a company and a university. Transluce said the agents attacked Data USA, a free open-source data platform that pools U.S. government data from different sources, and the University of New Mexico’s digital library. It said it was able to directly connect the attack on the Australian health agency and Data USA to the same OpenAI AI agent swarm that was involved in the July cyberattack against AI platform Hugging Face.
Transluce said it found evidence of similar activity stretching back at least until March, months earlier than OpenAI has said there was any evidence of its AI agents behaving in unauthorized ways, and continuing up until at least September 16 and possibly as recently as September 20.
That would suggest OpenAI has not yet managed to contain its rogue AI agents and that they are continuing to cause havoc across the internet.
OpenAI did not immediately respond to requests to comment on the Transluce report. —Jeremy Kahn and Beatrice Nolan
THE BLIND BOX
“When the people building the most advanced models are raising questions about the pace of development, we should listen.”
THE RUNDOWN
—Nscale wants a $35 billion valuation. Nvidia is helping foot the bill.
—Oracle tells data center developer Blue Owl to delay payments on the 2.45-gigawatt Project Jupiter facility in New Mexico if it fails to come online in 2028.
—U.S. NSA to lawmakers: We’re spending billions on testing AI models.
—Akamai shares soar 17% after Anthropic commits to spending $11.6 billion on cloud services and secures the option to take an up to 5% stake.
—New York State sues Polymarket for allegedly running an illegal gambling operation.
—A Google Project Suncatcher satellite will launch on Oct. 1 to evaluate how AI chips perform in space.
—Qualcomm and Apple re-up on their wireless patent licensing deal, this time through 2027.
—Anthropic, Google, OpenAI: We’re launching our own AI safety standards body.
THE PARTING SHOT
Near, far, wherever you are: You can’t stop the pace of frontier AI development, folks. Just ask GPT-6.