This is today’s edition of The Download, a weekday newsletter delivering the latest news and insights from the world of technology.
What OpenAI’s latest controversy means for the future of mathematics
OpenAI says one of its AI agents has solved one of mathematics’ most important unsolved problems. Under ordinary circumstances, the claim would represent a major scientific milestone. Instead, the announcement has been overshadowed by accusations that OpenAI failed to credit researchers whose AI-assisted work influenced the company’s solution.
Regardless of how those allegations are resolved, the episode could mark a turning point for mathematical research. AI systems increasingly appear capable of helping researchers tackle some of the field’s most difficult problems—but doing so may require computing power, funding, and expertise available only to a small number of frontier AI companies.
If that becomes the future of mathematics, human mathematicians will need to rethink how they work, collaborate, and receive credit for discovery.
Read on to learn how artificial intelligence could reshape mathematics.
—Grace Huckins
US battery storage installations reach a new record
Battery storage installations in the United States reached a new record in the second quarter of 2026, with 20.2 gigawatt-hours of new capacity coming online. That is enough electricity to meet the daily needs of approximately 600,000 households.
The surge highlights the urgent demand for lower-cost batteries and expanded energy storage as renewable power continues to grow across the US grid. The market is also evolving differently for utility-scale battery systems and residential energy storage.
Here’s a closer look at the forces reshaping the US battery storage industry.
—Casey Crownhart
This entrepreneur is building an AI agent that can plan ahead
Danihal Hafner’s office in San Francisco’s SoMa neighborhood is largely empty, but robots fill the space. Humanoid machines in different shapes and sizes hang like marionettes from racks running through the center of the expansive facility.
Hafner is not yet sharing many details about his new venture. He describes it as the next stage of a long-running effort to help AI systems navigate environments they have never encountered during training.
The 31-year-old researcher developed his approach by testing agents trained inside his world model against popular video games. More recently, researchers have begun moving these agents from virtual environments into the physical world.
Learn more about how Hafner is teaching AI systems to understand and navigate the world around them.
—Matt Honan
Danihal Hafner is one of the artificial intelligence innovators named to the 2026 list of 35 innovators under 35. Discover the rest of the list or explore the complete selection across biotechnology, AI, computing and robotics, climate, and energy.
MIT Technology Review Narration: Data from Ukraine’s drones is creating new Western defense markets
Drone debris covers battlefields across Ukraine. Behind that wreckage, however, lies a new defense resource: the vast amount of data generated by drone operations.
Ukraine has begun making millions of data points collected during tens of thousands of drone flights available to military contractors and commercial companies. The strategy offers a fast way to attract funding and partnerships, but it also turns the front lines into a unique AI training environment—one shaped by the chaos and complexity of war.
As this emerging industry develops, regulators will need to ensure that battlefield data is not treated like ordinary commercial information.
This is our latest article—a weekly narrated podcast from MIT Technology Review. Listen on Spotify or Apple Podcasts. Follow MIT Technology Review Narrated on your preferred platform to receive new episodes as they are released.
Must-read technology news
We searched the internet for the most important, surprising, alarming, and fascinating technology stories of the day.
1. OpenAI says it solved a 90-year-old mathematics problem in 88 hours
OpenAI used 10,000 AI agents to work on the Navier–Stokes equations. (CNBC)
+ OpenAI says the achievement is the first major mathematical problem solved by AI. (Nature)
+ The claim has been overshadowed by a dispute over research credit. (Axios)
+ OpenAI reportedly spent millions of dollars pursuing a million-dollar mathematics prize. (Quanta)
2. US accuses six Chinese AI companies of “industrial-scale” technology theft
The companies named include DeepSeek, Moonshot AI, Alibaba, and Z.AI. (CNN)
+ US authorities allege that the companies stole American AI trade secrets. (Reuters $)
+ The companies are alleged to have used model distillation to train their systems. (WSJ $)
+ The alleged targets include Claude, ChatGPT, Gemini, and Grok. (NBC News)
3. The Department of Defense asked OpenAI for AI models and rarely received a refusal
The military sought to “minimize the rejection rate.” (The Intercept)
+ The Pentagon says US allies are struggling to keep pace with AI advances. (The Guardian)
+ AI companies could soon train models using classified military data. (MIT Technology Review)
4. Apple is expected to unveil a $2,000 foldable iPhone
The device would represent the iPhone’s biggest design change since its 2007 launch. (The Guardian)
+ The launch would also be the first major test for new CEO John Tarnas. (The New York Times $)
+ Xiaomi and Huawei introduced new foldable phones ahead of Apple’s event. (CNBC)
5. Meta’s new AI agent can use apps to send emails and make payments
Meta’s Muse agent can autonomously interact with apps and websites on behalf of users. (CNBC)
+ Internal testing suggests sensitive personal information could be exposed. (Reuters $)
+ AI agents are not necessarily the same as human “colleagues.” (MIT Technology Review)
6. Google says complying with EU rules could “degrade” search results in Europe
The revised results may increase the visibility of comparison-shopping websites. (Reuters $)
+ The changes follow a €460 million European Union antitrust fine. (Quartz)
7. Meta ads promoted AI apps that digitally undress real teenagers
Researchers identified 332 advertisements containing child sexual abuse material this year. (BBC)
+ Investigators found photographs of real children that had been manipulated using AI. (Ars Technica)
+ Apple and Google missed a UK deadline to block child nudity on their phones. (Wired $)
8. Border Patrol is using financial data to target Americans during stops
Predictive-policing programs are providing information to local law-enforcement agencies. (404 Media)
9. New reflective paints could cool buildings without electricity
The paints reflect sunlight and release heat into space, offering a potentially low-cost way to cool buildings. (The Economist $)
10. A creepy first trailer for the Sam Altman biopic has been released
Luca Guadagnino’s Artificial, starring Andrew Garfield, is scheduled for release in the US on December 25. (Variety)
+ Amazon reportedly stopped releasing movies after investing in OpenAI. (The Guardian)
Quote of the day
“This is a Deep Blue and Kasparov moment. The community needs to have a serious and urgent discussion about where to go from here.”
—New York University mathematician Tristan Buckmaster, in a statement comparing OpenAI’s mathematical advances with IBM’s historic defeat of chess champion Garry Kasparov in 1997.
One more thing
Shocked to discover my body had been used in deepfake pornography
When Jennifer began a research position in 2023, she ran a professional headshot through a facial-recognition program. She wanted to find out whether pornography she had made more than a decade earlier could still be located online. It could—but the search also revealed something she had never seen before: one of her old videos had been digitally altered so that another person’s face appeared on her body.
Discussions about sexual deepfakes often focus on people whose faces are inserted into explicit content without consent. Another group is frequently overlooked: people whose faces are removed from their bodies and replaced with someone else’s likeness.
Adult-content creators say AI systems are training on their work, cloning their likenesses, and generating explicit material they never agreed to produce. They have few legal protections and limited regulatory support. Read the full story about the threats AI deepfakes pose to creators’ rights, livelihoods, privacy, and control over their own bodies.
—Jessica Klein
We still have good things
A place for comfort, fun, and distraction to brighten your day. Have an idea? Please contact me.
+ Workers are finding ways to avoid and challenge employers’ AI obligations.
+ Artist Ali Hill creates remarkable embroidered buildings and streets.
+ Explore the extraordinary anatomical structures that allow elephants to hear vibrations through their bones—possibly helping them communicate across more than six miles.
+ View the breathtaking winning images from the 2026 International Aerial Photographer of the Year competition.
Source: www.technologyreview.com