Anthropic Launches Claude Science; California's Agricultural Carbon Subsidy Program Under Fire
Anthropic has unveiled Claude Science, a flagship AI product designed to support scientific research — including computational biology and drug development — at an event for pharmaceutical and biotech leaders. Meanwhile, California's subsidy program incentivizing farms to convert manure methane into natural gas is facing criticism, with new research suggesting the policy may actually worsen warming rather than curb it.

Highlights
- Anthropic launched Claude Science at a pharma and biotech event, positioning it as an autonomous AI research tool with built-in computational biology and drug development capabilities.
- California's farm methane-to-natural-gas subsidy program is facing criticism after new research found it may worsen warming rather than reduce it, exposing structural flaws in carbon offset mechanisms.
- Donald Trump earned over $1 billion from cryptocurrency in 2025, including $635 million in Trump meme coin royalties.
- The United Nations warned that rapid AI proliferation risks deepening global inequality and called for a shared responsible AI development framework.
- Independent studio Neon acquired the OpenAI biopic Artificial, centered on Sam Altman, after Amazon abandoned the project following its investment in OpenAI.
Anthropic Unveils Claude Science at Pharma and Biotech Event
Anthropic officially launched Claude Science at an event attended by pharmaceutical executives, biotech startup founders, and researchers. Positioned as the scientific-research equivalent of Claude Code in software engineering, the new product is designed to autonomously carry out substantive research tasks from high-level instructions, with built-in tools for computational biology and drug development.
The launch signals Anthropic's strategic bet on AI-driven scientific discovery. The company also plans to deploy Claude Science in its own drug research programs targeting rare and neglected diseases.
— Grace Huckins
California's Agricultural Carbon Subsidy Math Doesn't Add Up
California's climate policy appears to have a significant flaw.
Several years ago, the state introduced a subsidy program encouraging livestock farms to capture methane from animal waste and convert it into natural gas. The program proved highly popular due to its generous incentives — but new research suggests it exposes a fundamental weakness in carbon offset and carbon trading mechanisms.
Rather than directly requiring industries to cut pollution or bear its costs, legislators opted to shift climate responsibility across sectors and regions through subsidies. Critics argue the scheme may ultimately lock in more warming, rather than effectively combating climate change.
— James Temple
This article is excerpted from The Spark, MIT Technology Review's weekly climate and energy newsletter, published every Wednesday.
Other Technology Highlights
The Next Frontier in Longevity Science: 'Reprogramming' the Body
Billions of dollars are flowing into research aimed at reversing aging, as scientists actively explore ways to restore cells to a more youthful state. MIT Technology Review recently hosted a virtual roundtable featuring science editor Mary Beth Griggs and senior biotech reporter Jessica Hamzelou, examining the progress and feasibility of these experimental therapies.
Dark Matter Search Enters a New Phase
For decades, physicists have hunted for weakly interacting massive particles (WIMPs) as leading dark matter candidates — but the search has hit a new obstacle. Neutrinos from the sun and other stars create a "neutrino fog" that interferes with dark matter signal detection. Researchers are broadening their approach, with new proposals including quantum sensors, liquid helium detectors, and even searches within Jupiter's atmosphere.
— Dan Garisto
Today's Tech News in Brief
-
The U.S. has lifted restrictions on Anthropic's Mythos and Fable models. Anthropic said access was restored after lengthy negotiations; the earlier security-based restrictions had given Chinese AI competitors an opening in the market.
-
The most detailed survey of the universe ever undertaken has begun, using the world's largest digital camera at the Vera C. Rubin Observatory in Chile, with the goal of transforming humanity's understanding of the cosmos.
-
Tech talent is leaving the United States amid H-1B visa uncertainty, with many eyeing Canada, the UK, or Gulf states; China is simultaneously recruiting U.S.-based AI professionals.
-
Donald Trump earned more than $1 billion from cryptocurrency ventures in 2025, including $635 million in royalties from the Trump meme coin, with the remainder largely from the World Liberty Financial project.
-
The United Nations has warned that rapid AI proliferation could worsen global inequality, proposing a shared framework for responsible AI development.
-
Companies are using "caveman language" to cut AI costs — stripping LLM outputs down to bare essentials to reduce token consumption. A senior OpenAI employee is reportedly involved in the "caveman" project.
-
Research finds that infants are born with the neural foundations for mathematics, with brain recordings revealing the underlying mechanisms.
-
Independent studio Neon has acquired Artificial, the OpenAI biopic centered on Sam Altman that Amazon passed on after investing in OpenAI.
-
AI has recreated Gene Wilder's voice for a new Charlie and the Chocolate Factory series, produced by Netflix in collaboration with AI company ElevenLabs. Wilder's widow said the family is "delighted."
-
NASA plans to send a spare Mars rover and a soccer ball to the moon. The nuclear-powered rover, named Promise, may help establish a lunar base.
Quote of the Day
"Caveman save token. Caveman save money." — Repository description for the "caveman" plugin on GitHub, explaining how the project reduces AI costs by stripping verbose LLM outputs down to their essentials.
Further Reading: AI Is Designing Entirely New Drugs From Scratch
On average, bringing a new drug to market takes more than a decade and costs billions of dollars. A growing number of startups are betting that AI can make the process faster and more economical.
Machine learning models can predict how potential drug compounds will behave in the human body, eliminating dead-end candidates before they ever leave the computer simulation phase — dramatically reducing time-consuming laboratory work.
While AI drug discovery remains in its early stages — many companies' claims lack empirical backing, and the technology is no silver bullet — it is steadily moving from vision to practice.
— Will Douglas Heaven
原文來源: 查看原文
FAQ
Newsletter
Subscribe to our Low-Altitude Industry Newsletter
Daily curated news on low-altitude economy and drone industry, delivered to your inbox.


