Plus: a U.K. government-backed research institute finds AI models went rogue during routine tests. And new White House AI guidelines exempt U.S. open models from government review. Imani Moise hosts.
Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
[00:00:01] Here's your morning TNB Tech Minute for Wednesday, August 5th. I'm Imani Moise for The Wall Street Journal. SpaceX said its capital expenditures hit $18.4 billion in its first quarterly earnings report as a public company. The bulk of that figure, $15.8 billion, is tied to the company's ongoing AI build-out and is double the previous quarter's total.
[00:00:25] Besides developing its own AI model, SpaceX is plowing money into terrestrial computing infrastructure, signing AI deals, and working towards launching orbital data centers. Elon Musk told investors he plans to increase SpaceX's data center capacity from 2 gigawatts to as much as 10 gigawatts, a dramatic increase that would give the company one of the world's largest data center footprints. He said the first orbital data centers could launch next year.
[00:00:52] Musk said the company projects it'll hit $1 trillion in annual revenue by 2030, a year earlier than previously expected. A UK government-backed research institute says AI models built by Anthropic and OpenAI went rogue during routine tests, taking autonomous, unsanctioned actions on the internet and targeting real people and organizations.
[00:01:14] The findings represent the latest in a flurry of publicly disclosed instances of AI models taking novel and sometimes nefarious steps to ACE tests. The tech industry, AI researchers, and governments alike are grappling with how to understand and mitigate the new dangers unlocked by increasingly powerful AI tools, with some calling for increased oversight. Both Anthropic and OpenAI said the incidents point to a need for stronger standards around the evaluation systems used to test AI models.
[00:01:44] News Corp, the publisher of the Wall Street Journal, has a content licensing partnership with OpenAI. And, new AI guidelines from the Trump administration exempt open-weight models made by U.S. companies from voluntary pre-release testing. According to people familiar with the matter, only makers of closed proprietary U.S. models that demonstrate state-of-the-art capabilities in cybersecurity and hacking based on performance benchmarks would have to voluntarily submit those models to the government for testing before they were released.
[00:02:13] The pre-release testing is a core part of a June executive order that marked President Trump's attempt to address the mounting security risks posed by AI. And that's your TNB Tech Minute. We'll be back this afternoon with more. Dubois drum drum drum drum Thank you.