The Download: AI agents for science, and the “censorship-industrial complex”

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.

AI for science needs reasoning, not just data

—Eric Schmidt, the former CEO of Google and the cofounder of Schmidt Sciences, and Suhas Mahesh, who leads the AI for science work at the AI Center of Schmidt Sciences

In 2024, Google DeepMind scientists shared the Nobel Prize in Chemistry for a neural network, AlphaFold, which predicts the structures of proteins. It showed that AI could make groundbreaking scientific discoveries, but AlphaFold may not be the best template for accelerating science. Instead, another approach may hold the key: AI agents.

AlphaFold relied on a dataset of roughly 170,000 experimentally validated protein structures that took 53 years and roughly $21 billion worth of experimental work to assemble. Comparable datasets will be difficult or impossible to create in many fields.

AI agents can instead model the iterative, highly contingent process of actual research. While tools like AlphaFold apply a powerful approach to a limited question, agents are inherently generalists. They do not represent a new way to do science—instead, they digitally model the human process of discovery.

Read the full op-ed on how agents could accelerate scientific discovery.

Inside the “censorship-industrial complex” idea shaping US policy

For years, the idea of a “censorship-industrial complex” that suppressed conservative and populist speech spread in right-wing circles online. But now, the theory has made its way into the Trump administration.

Over the past nine months, MIT Technology Review investigated its origins and traced its rise. In a virtual Roundtables session on Thursday, August 13, senior reporter Eileen Guo and executive editor Amy Nordrum will explore what they discovered, where the theory is going, and what it could mean for the future of democracy and the internet.

Register here to join the session at 19:00 GMT / 2:00 pm ET / 11:00 am PT.

The must-reads

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

1 A new Amazon data center could become the US’s biggest polluter The Texas facility has a permit to release up to 33 million tons of CO2. (NYT $) + It will be Amazon’s first off-grid AI data center. (Cleanview) + The gas plant will generate up to 7.65 gigawatts of power. (Verge)

2 OpenAI has paused work on its Astra AI model over security concerns Tests found it could launch cyber-attacks autonomously. (FT $) + Critics warn such disclosures could be designed to spark hype. (Guardian) + AI doesn’t need to be that smart to make online crimes easier. (MIT Technology Review)

3 North Korean hackers have built AI tools for cyberattacks A new report says they’re used in spear-phishing campaigns. (Al Jazeera) + They’re attributed to the state-linked group Kimsuky. (Reuters $)

• Which could automate attacks and analyse stolen data. (TNW)

  4 China’s firms control 97% of global humanoid shipments Global shipments have more than tripled since last year. (Bloomberg $)** ** *+ Chinese humanoid manufacturers are racing to public listings.* ([Reuters](https://www.reuters.com/world/asia-pacific/what-is-unitree-why-are-chinas-humanoid-robot-makers-racing-list-2026-08-10/)$) + Gig workers are training humanoids at home. (MIT Technology Review)

  5 Taiwan has expanded its use of aerial and maritime drones to deter China The approach draws on lessons from Ukraine’s battlefield. (NYT $) + Taiwan’s “silicon shield” could be weakening. (MIT Technology Review)

  6 Apple is testing Chinese memory chips as the supply squeeze bites But any deal with CXMT would be politically sensitive. (WSJ $)

  7 An Israeli startup is tied to rogue AI hacks at OpenAI, Anthropic, and Meta Irregular’s tests allowed the models to access the public internet. (CNBC)

  8 NASA is flying into wildfire storms to understand them The mission will study their effects on fires and the atmosphere. (Economist $)

  9 A chatbot-free childhood may become a status symbol AI is like ultra-processed food for developing brains. (Atlantic $) + Are chatbots making us lose control of our brains? (MIT Technology Review)

  10 MySpace is eyeing a comeback as an “antidote” to social media fatigue But the nostalgic plan looks like a long shot. (BBC)

  Quote of the day

  “It’s the humans that we need to watch out for. AI is just the tool.”

  —Oren Etzioni, professor emeritus at the University of Washington and former CEO of the Allen Institute for AI, tells CNN that people are still the main threats in cybersecurity.

  One More Thing

  图片

  Meet the man building a starter kit for civilization

  Marcin Jakubowski lives in a house he designed and built himself. He relies on the sun for power, heats his home with a woodstove, and farms his own fish and vegetables.

  Jakubowski is the founder of Open Source Ecology, a collection of 50 machines capable of building civilization from scratch. It includes everything from a tractor to an oven to a circuit maker, all designed to be reconfigured however you see fit.

  The ultimate goal is a “zero marginal cost” society, where producing an additional unit of a good or service costs little to nothing. Jakubowski hopes to get there by eradicating licensing fees, decentralizing manufacturing, and fostering collaboration through education.

  Take a look inside his starter kit for civilization.

  —Tiffany Ng

  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.)

  • A rhythmic doggie is proving that not only humans can play the drums.
• Catch a glimpse of the internet’s future at the Awwwards, the Oscars of web design.
• Explore the ocean year-round with this immersive experience of the Nautilus vessel’s expeditions.
• Earth once (or twice) had supermountains. This fascinating video explores their links to evolutionary shifts in the history of life.