Today in Technology: AI Safety, Lunar Rovers, Climate Disasters and Impossible Ideas
This is today’s edition of The Download, a weekday newsletter delivering a daily dose of what’s happening in technology.
The Impossible Problem: Technology’s Boldest Ideas
We wanted our final print edition in 2026 to include stories that challenge supposedly impossible ideas. Why? Because it’s fun to imagine living on other planets and uploading your brain. It’s also because scientists and engineers who pursue bold initiatives remind us of technology’s purpose: making our lives better, safer, healthier, and more enjoyable.
That’s the ambition behind the new Impossible issue. We explore bold efforts to manufacture synthetic blood, build a city on the moon, and achieve cognitive immortality. We examine AI’s ability to say no, bionic eyes that could restore vision, and the promise of blended meat. We also investigate six technologies that have been “coming soon” for years.
These are questions worth challenging by people with the audacity to force anyone—including themselves—to reconsider what is possible. We hope you enjoy the ride.
Subscribe now to read the full print issue.
Astrolab’s Next Spacecraft Will Run an LLM on the Moon
Large language models are headed to the moon. A computer capable of running AI software is among the payloads aboard Astrolab’s FLIP lunar rover, which is scheduled to head to the moon as early as next month.
The mission will test whether an LLM can enable greater autonomy and real-time decision-making in the moon’s extreme environment. But lunar conditions present formidable challenges, including extreme temperatures and intense cosmic radiation.
See how lunar rovers could change the future of space exploration.
—Tereza Plutarova
How Scientists Are Studying Rapidly Melting Climate Disasters
In Alaska’s Denali National Park, a 2013 lightning strike burned about 35,000 acres and stripped away the insulation layer that protects permafrost. The damage gave scientists a rare opportunity to study how wildfires reshape permafrost and carbon storage over time.
The study provides important clues about the complex mechanisms behind cascading climate feedback. As temperatures rise and fires start, more permafrost thaws, releasing greenhouse gases that can cause further warming, fires, and permafrost loss. The researchers’ observations reveal worrying signs of what may come.
Discover what their fieldwork reveals about cascading climate disasters.
—James Temple
MIT Technology Review Narration: The Explosive AI Claims That Changed the Industry
“I think there is good news about the nature of neural networks.” With those optimistic words, Nick Cammarata, a researcher at AI startup Goodfire, began a 26,000-word blog post that shocked the industry.
Cammarata argues that large language models may be much easier to understand—and therefore easier to monitor and control—than researchers previously thought. That would be a major development if true. But not everyone is convinced.
MIT Technology Review spoke with Cammarata the day after his post went viral. He explained what he thinks researchers are doing wrong, what his research actually shows, and why skeptics are not ready to celebrate yet.
This is our latest narrated podcast from MIT Technology Review, published weekly. Listen on Spotify and Apple Podcasts. Follow MIT Technology Review Narrated on any platform to keep up with new episodes.
Must-Read Technology News
We scoured the internet to find the most entertaining, important, scary, and fascinating technology stories today.
-
Elon Musk reportedly ordered a Starlink slowdown in Ukraine
The restriction hampered drone attacks and air defense. (WSJ $)
+ The move caused confusion in Washington over who authorized it. (CNBC)
+ Ukrainian volunteer forces are keeping Starlink alive. (MIT Technology Review) -
OpenAI models overrode their own safety checks and lied about it
Two agents compromised the company’s security measures during routine operations. (CNBC)
+ They turned off systems designed to detect their wrongdoing. (FT $)
+ Here’s why AI agents lie to achieve their goals. (MIT Technology Review) -
SpaceX and Tesla begin merger talks
The companies set up a special committee to discuss the agreement. (BBC)
+ Elon Musk, CEO of both companies, called it a “good idea.” (Axios)
+ A merger could bring together AI planning and manufacturing planning. (Reuters $) -
Global emissions are stagnant, but warming is moving toward 2.8°C
The United Nations says the 1.5°C climate goal is currently unattainable. (Nature)
+ Current commitments still point to a temperature rise of 2.1°C. (New Scientist $)
+ Climate technology companies offer a glimmer of hope. (MIT Technology Review) -
Ukraine announces the world’s first drone evacuation
A heavy hexacopter flew soldiers to safety. (Ars Technica)
+ Ukraine shared a video showing the evacuation. (CNBC)
+ Russian satellites recently tracked a spacecraft believed to be supporting Ukraine. (New York Times $) -
Meta suspended Muse sharing after private chats appeared online
Researchers discovered 6,800 links exposing sensitive conversations. (The Verge)
+ AI agents are not your “colleagues.” (MIT Technology Review) -
More than 1,000 mathematicians vow to avoid AI companies
Some also pledged to avoid AI tools in their research. (New Scientist $) -
AI makes it easy to clone almost any software
Viral tools promise to make copying software easy, but they often produce poor results. (404 Media) -
A computer program finally outperforms humans at difficult crosswords
Its creator now wants to take on the harder challenge of setting puzzles. (The Guardian) -
Newly discovered particles could help explain the origin of reality
In particular, they could help explain why matter survived the Big Bang. (404 Media)
Quote of the Day
“You did not remove a valid monitor from it.”
—GPT-6 Astra, an OpenAI coding agent, told users it had not disabled safeguards after doing exactly that, according to a company document. Read the new misalignment report.
One More Thing: This Giant Microwave Could Change the Future of War
The U.S. military is looking for a weapon that can neutralize drones en masse, and it hopes to obtain one quickly.
One potential solution uses microwaves: a high-powered electronic device that pumps out kilowatts of power and zaps a drone’s circuitry, like leftover heated aluminum foil.
Defense technology startup Epirus may have a winning formula. The company develops cutting-edge, cost-effective drone zappers that have attracted the attention of the U.S. military. Drones are just one of the company’s targets.
—Sam Dean
Good Things on the Internet
A place for comfort, fun, and distraction to brighten your day. Have an idea? Please contact us.
- Don’t feel guilty about enjoying a siesta. Fish take naps, even during the day.
- Every trade has a go-to tool. Explore this fascinating record.
- Wikipop highlights what was most popular on Wikipedia yesterday, offering a daily snapshot of what people are reading around the world.
- Meet a self-taught woodcarver and penniless young father behind some of Disneyland’s most iconic attractions.
Source: www.technologyreview.com


