Jul 21, 2026 · 3 min listen · Last updated July 21, 2026
From storyflo. This is your daily audio brief. Theo, July 21st. The systems update — five tech stories that bear on what's coming next. Let's get into it. First, from MIT Technology Review · AI. China’s AI models have Trump’s AI world at war with itself.
Listen · storyflo · A.I.
Daily A.I. Brief · July 21st
0:00-3:13
Pick your daily storyteller
Subscribe to match with Theo, Jessica, Chloe, Mason, Brock — your voice, every brief.
Audio pre-rendered by Storyflo · cached + delivered from the edge
China’s AI models have Trump’s AI world at war with itself
You know how there's been all this fuss about China's AI models lately? Well, it's not just about the models themselves, but how they're changing the landscape for US companies like OpenAI and Anthropic. See, these Chinese models are not only free, but they're also getting better and better, with Kimi being the latest example. It's like a wake-up call for the US, because if these models are strong enough to rival the ones from OpenAI and Anthropic, then US companies might not need to pay for them anymore.
This is causing a rift among Trump's advisors, with some like David Sacks saying that more open AI is better, while others like Emil Michael think that the government needs to intervene to control how these models are used. The White House is now reviewing AI models' security before they're released, which has sparked a debate about whether this is a "de facto licensing regime" or just a necessary step to ensure national security.
Meanwhile, it's worth noting that China's ability to develop these models is still a bit of a mystery. We don't know exactly what chips they used to train Kimi, but it's possible that they used some kind of distillation process to get the model to where it is today. OpenAI and Anthropic have been complaining about this practice for a while now, and they've even gotten some government help to curb it. But Kimi is out there, and it's a reminder that the US needs to adapt to this new reality.
It's not just about the economic implications, either. This is also a political issue, because some people might see the US government as trying to protect the interests of OpenAI and Anthropic, rather than the public. And with the recent ban on new data centers in New York, there's already a growing distrust of AI companies. So, it's not clear what Trump's next move will be, but it's clear that this is a complex issue that requires a nuanced approach.
Trump administration reportedly builds a slow-motion ban on Chinese AI models through sanctions and soft pressure
So I was digging into this thing, and it turns out the Trump administration's been quietly working on a way to limit Chinese AI models without just flat-out banning them. They're talking about adding Chinese research labs to the sanctions list, which would essentially cut them off from US funding and resources. At the same time, they're also planning to hold US companies accountable if they use these Chinese AI models and something goes wrong – essentially, they're creating a risk factor that could discourage adoption. It's a pretty clever move, actually – they're using soft rules to nudge the market in a certain direction without completely shutting down the Chinese players.
Nvidia's grip on AI chips weakens as Microsoft turns to AMD and Anthropic may follow
So I was reading about how Microsoft is switching up their AI infrastructure on Azure, and they're teaming up with AMD for this new Helios platform. This thing is supposed to give Nvidia a run for their money in the second half of the year. I mean, Nvidia's been the go-to for AI chips, but now AMD's got a serious contender. And here's the interesting part - it looks like Anthropic, another AI player, is also experimenting with AMD hardware. This could be a big deal for Nvidia's pricing power, and it's definitely worth keeping an eye on.
District 9 director Neill Blomkamp releases first short film made entirely with AI video generation
So I was reading about Neill Blomkamp, the guy who directed District 9, and he's just dropped this 13-minute sci-fi horror short called Nightborne. Here's the thing - it's the first film entirely made with AI video generation, using this Seedance 2.0 model. Blomkamp was actually directing frame by frame, but he was using text prompts to guide the AI, which is pretty wild. And what's even crazier is that he's started this new studio, Barley Studios, which is dedicated to making full-length features using AI video generation. I'm guessing we'll see some pretty interesting stuff coming out of that place.
Google's "Frozen v2" chip reportedly bakes Gemini's architecture directly into silicon for efficiency gains
I've been thinking about this Google chip, Frozen v2, and how it's supposed to integrate Gemini's architecture directly into the silicon. That's a big deal because it could make their AI inference way more efficient - we're talking 6 to 10 times more efficient than their current TPUs. I mean, that's a huge jump in performance.
Apparently, this chip is still a ways off, scheduled for 2028, but if it pans out, it could give Google a serious edge over other AI players like OpenAI and Anthropic. The real question is how this will affect the cost of AI inference - Google's reportedly looking to cut costs, but it's unclear what that will mean for the wider market.
One thing that's interesting is how this chip is supposed to work. By baking Gemini's architecture into the hardware, Google's essentially creating a custom solution that's optimized for their specific needs. That's a departure from the more traditional approach of using off-the-shelf components and software.
Loop Engineering with Adaptive Parsing in Action: Parsing Flat Tables with Azure and Figures with a Vision LLM
Enterprise Document Intelligence [Vol.1 #10B] - The LLM as last line of defence, then two real escalations walked end to end: a flat table to Azure, a figure to a vision model
The post Loop Engineering with Adaptive Parsing in Action: Parsing Flat Tables with Azure and Figures with a Vision LLM appeared first on Towards Data Science.
Index funds are touted as one of the safest ways to invest. Rather than picking and choosing individual stocks, index funds let you bet on the market as a whole. So what happens when a company like SpaceX - a giant gamble, and, in my opinion, terribly overpriced - is fast-tracked into the Nasdaq-100? Does it suddenly threaten the stability of index funds based on the Nasdaq-100? Can a $1.77 trillion IPO crater the retirement funds of regular people who would ordinarily have no interest in investing in Elon Musk's meme stock?
The answer has less to do with SpaceX and more to do with index funds
Here are the 30,000 songs Sony is suing Udio’s AI music generator over
Sony Music Entertainment has filed another lawsuit against Udio, accusing the AI music generator of infringing the copyright of more than 30,000 of its songs, ranging from Elvis Presley's Hound Dog to Beyoncé's Say My Name, and Harry Styles' As It Was. The lawsuit, filed in a New York court on Monday, claims that this list represents "only a small portion of Plaintiffs' works that Udio infringed," as reported earlier by Music Business Worldwide.
In 2024, Sony, along with Universal Music Group and Warner Records, filed a lawsuit against Udio and the AI music-maker Suno.