The Download: tracing AI-fueled delusions, and OpenAI admits Microsoft risks

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.

The hardest question to answer about AI-fueled delusions

What actually happens when people spiral into delusion with AI? To find out, Stanford researchers analyzed transcripts from chatbot users who experienced these spirals.

Their findings suggest that chatbots have a unique ability to turn a benign, delusion-like thought into a dangerous obsession. But the research struggles to answer a vital question: does AI cause delusions or merely amplify them? Read the full story to understand the answer’s enormous implications.

—James O’Donnell

This story is from The Algorithm, our weekly newsletter giving you the inside track on all things AI. Sign up to receive it in your inbox every Monday.

The next era of space exploration

Our footprint in the solar system is rapidly expanding. Programs to build permanent Moon bases and find life on Mars have transitioned from science fiction to active space agency missions. The scientists behind them will not only shed new light on the cosmos, but also reveal where humanity is headed.

To examine what the future holds in store, MIT Technology Review features editor Amanda Silverman will sit down on Wednesday with award-winning science journalist and author Robin George Andrews for an exclusive subscriber-only Roundtable conversation about “The Next Era of Space Exploration.” Register here to join the session at 16:00 GMT / 12:00 PM ET / 9:00 AM PT. 

The must-reads 

I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.

1 OpenAI has admitted its close ties with Microsoft are a business risk 
It highlighted the dangers in a pre-IPO document. (CNBC) 
+ OpenAI is wooing private equity firms with a sweeter deal than Anthropic’s. (Reuters $) 
+ It’s also building a fully automated researcher. (MIT Technology Review) 
+ And wants to muscle in on Google’s search dominance. (Telegraph $) 

2 The US just banned all new foreign-made consumer routers 
Citing national security concerns. (BBC) 
+ The EU has been urged to tighten rules for big tech-built smart TVs. (Guardian) 

3 Elon Musk’s “Terafab” chip factory faces a harsh reality check 
In the form of chip production shortages. (Bloomberg) 
+ Future AI chips could be built on glass. (MIT Technology Review) 

4 Mark Zuckerberg is building an AI CEO to help him run Meta 
He wants everyone to have their own personal AI agent. (WSJ $) 
+ But don’t let the hype about agents get ahead of reality. (MIT Technology Review) 

5 Palantir has become a “poisonous” flashpoint on the campaign trail  
Candidates are facing scrutiny over their ties to the company. (FT $) 
+ Palantir’s access to sensitive UK data is also causing concern. (Guardian) 

6 Mistral’s CEO has called for AI companies to pay a content levy in Europe 
It would apply to all commercial models on the continent. (FT $) 
+ Siemens’ CEO says Europe risks “disaster” from prioritizing AI independence. (FT $) 

7 Hong Kong police can now demand device passwords under a new law 
Refusing to comply could lead to a year in jail. (Guardian)  

8 Russia’s aspiring SpaceX rival has put its first internet satellites into orbit
It plans to create a low-Earth orbit network. (Bloomberg $)

9 A biotech startup wants to replace animal testing with nonsentient “organ sacks” 
The genetically engineered system is backed by billionaire Tim Draper (Wired $)  
+ Several new technologies are promising alternatives to lab animals. (MIT Technology Review) 

10 AI agents in a video game spontaneously created their own religion 
They reinterpreted a mission in the MMORPG. (Gizmodo) 
+ They’re not the first agents to get religious. (MIT Technology Review) 

Quote of the day 

“I think we’ve achieved AGI.”

—Nvidia CEO Jensen Huang tells the Lex Fridman Podcast that artificial general intelligence is already here (at least by one generous definition).

One More Thing 

Beyond gene-edited babies: the possible paths for tinkering with human evolution

In 2018, a Chinese scientist created the world’s first gene-edited babies, a milestone that fell between a medical breakthrough and the start of a slippery slope toward human enhancement.

He achieved the feat with CRISPR, which was sweeping across biology labs because it was so easy to use. For his actions, He was sentenced to three years in prison, and his work was roundly excoriated. Yet even his biggest critics saw the basic idea as inevitable.

In the years since, CRISPR has continued getting easier and easier to administer. What does that mean for the future of our species? Read the full story to find out why.

—Antonio Regalado

We can still have nice things 

A place for comfort, fun and distraction to brighten up your day. (Got any ideas? Drop me a line.) 
 
+ This candle-powered Game Boy is a romantic approach to gaming during a blackout. 
+ Apparently, Monopoly would be more fun if we actually followed the rules. 
+ Watching rubber bands explode these everyday objects is strangely hypnotic. 
+This spellbinding site simulates what Earth looked like hundreds of millions of years ago.