Hi and welcome to TechScape. Nick Robins-Early here, US tech and power reporter at the Guardian. I’m filling in for your usual host Blake Montgomery, who is out this week on vacation.
Today, we’ll be talking about the historic SpaceX IPO and the US government’s surprise order to limit the use of Anthropic’s most advanced AI model over cybersecurity concerns. I’ll also share a dispatch from Web Summit Rio, South America’s largest tech event.
SpaceX IPO mints Musk as a trillionaire
Elon Musk’s SpaceX hit the market on Friday in the biggest IPO of all time, raising $85.7bn and easily shattering the previous record of $29.4bn set by the Saudi oil giant Aramco. The rocket, AI and satellite communications company ended the day at $160.95 per share, up from its IPO price of $135 and satisfying any Wall Street skepticism over the unorthodox rollout of the stock.
SpaceX’s successful market debut turned Musk into the world’s first trillionaire, an unprecedented accumulation of wealth that supporters touted as a testament to his financial genius and critics denounced as a symbol of a broken economic system. Prominent Democrats including Elizabeth Warren and Alexandria Ocasio-Cortez issued posts on X denouncing Musk’s exorbitant fortune and calling for a wealth tax on the ultra-rich.
Musk’s new-found trillionaire status further cemented the immense amounts of wealth being consolidated in the tech boom, with Anthropic and OpenAI also set to hold blockbuster IPOs later this year at sky-high valuations. While putting unfathomable sums of money into the hands of major investors and tech moguls, these companies have also become load-bearing pillars of the US economy itself. Everyday Americans also are having their financial futures increasingly intertwined with companies like SpaceX through 401k retirement funds and index funds, putting everyone at risk should these firms struggle to meet their lofty goals.
Musk claimed on Sunday that SpaceX could bring in $1tn in revenue by 2030. Musk says a lot of things that do not pan out, and some major market analysts are far more skeptical about what kind of returns the company will see in the next five years. Much of it depends on what happens in the broader AI boom and whether the company’s plans for putting datacenters in space, among other moonshot ideas, ever come to pass. Now that SpaceX is public, it will be harder than ever not to have a stake in the outcome.
More on SpaceX’s IPO
-
SpaceX makes largest ever stock market debut, minting Musk as a trillionaire
-
After SpaceX’s huge IPO, Americans’ financial future will be bound to AI
-
How much money did Elon Musk make in SpaceX’s stock market debut?
Another front in the fight between Anthropic and the White House
The US government announced late on Friday that it was ordering Anthropic, the artificial intelligence company best known as the maker of Claude, to halt all foreign nationals from being able to access its most advanced AI models, known as Fable 5 and Mythos 5. The administration cited national security risks and concern that adversaries could bypass the models’ safety guardrails to use the AI tools for cyberattacks.
Anthropic said it would comply with the directive, although it disputed the government’s assertion that there was a general way to bypass, aka “jailbreak”, its models and use them for harm.
“We are complying with the government’s legal directive and are removing access to Fable 5 and Mythos 5 for all users,” Anthropic said in a statement. “However, we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people”.
The order reignites tensions between the Trump administration and Anthropic after what seemed like a period of detente. Both sides are still locked in a legal battle over Anthropic’s refusal earlier this year to remove some safeguards on its AI models built for government use over concern they could be used for mass surveillance or lethal fully autonomous warfare.
Anthropic’s CEO Dario Amodei has reportedly been on a number of harried calls with administration officials as they try to sort out the latest dispute, but the latest dust up is more evidence of the lack of a comprehensive framework for how to deal with the unexpected threats generated by AI. There is no consensus over exactly who gets to decide when, how and by whom new AI tools get utilized, which is an increasingly pressing question as these rapidly advancing models pose a deepfake and cybersecurity nightmare.
Read more: Anthropic disables its most advanced models after US order limiting foreign access
Who gets to decide what AI will sound like?
While hosting panel talks at the Web Summit Rio tech conference last week, I spoke with Rodrigo Durán, the executive director of Chile’s National Center for Artificial Intelligence. Durán has been working on building Latam-GPT, a regional effort to build an open source, free AI model that will cater specifically to the needs of Latin America.
The project has been a massive undertaking that has involved coordination across 15 countries and dozens of institutions, in addition to using eight terabytes of data for its training. The goal is to create a model that can understand and navigate the intricacies of Latin America’s many languages and dialects, as well as its historical and cultural nuances, in a way that Silicon Valley-based products like ChatGPT or Claude are not capable of doing because of built-in biases and limited data.
I found Latam-GPT a fascinating endeavor, not only because it is part of a growing movement surrounding digital sovereignty from US tech giants, but because of the questions it brings up about how AI models are shaping our cultural and linguistic future.
The growth of new technologies often impacts language, sometimes inventing new ways of speaking and in other cases developing a single standard which boxes out alternative voices. The BBC’s use of a specific “received pronunciation” accent in its early radio broadcasts, something that came to be known as “BBC English,” influenced the way that people learned English and thought about what constituted a “proper” British accent for generations.
Many people can now recognize when text or audio feels like it was AI-generated, as popular AI tools tend to create content with a certain uncanny tone and reliance on a few common sentence structures, words and punctuation choices. The decisions made by AI tools over how to speak, which accents or dialects to prioritize, and what counts as “neutral” language, could similarly shape the way we understand language in future generations.
What Durán is concerned about, and a question for everyone across continents, is who gets to have a say in the way we talk to AI, and AI talks to us.