AI Daily Briefing — 2026-07-11

On America’s 250th birthday, four nuclear microreactors achieved criticality—a technical milestone meaning they can sustain a chain reaction. This surpasses the goal set by the previous administration, which called for three reactors to reach this stage by July 4. The achievement comes at a time when demand for clean, emissions-free electricity is rising. However, reaching criticality does not mean these reactors are ready to supply power to the grid. The milestone signals progress for nuclear technology, but each company still faces significant steps before commercial operation, reports MIT Technology Review AI.

In other tech news, China is reportedly preparing to allow its leading AI firms—including Alibaba, ByteDance, and DeepSeek—to purchase Nvidia’s H200 chips. The U.S. had previously authorized these sales, but Beijing withheld approval. The move could boost China’s AI capabilities amid ongoing tensions over semiconductor exports. Meanwhile, NATO is building a detection network using sensors, drones, satellites, and AI to counter Russian cyberattacks. Researchers are also exploring controversial geoengineering methods, such as dimming sunlight, to mitigate future El Niño events, though experts warn of unintended consequences.

Meta is patenting an AI device that records users to analyze emotions, ostensibly for personalized workout plans, raising privacy concerns. Chipmakers are stacking transistors vertically to keep advancing as Moore’s Law slows. A study found that Ivy League students suspected of AI cheating saw their in-person test scores drop from 96% to 48%. Parents’ phone addictions are linked to weakened bonds with children, potentially causing lifelong insecure attachment. A judge approved Elon Musk’s $1.5 million Twitter settlement with the SEC despite citing “serious misgivings.” Finally, shoebox-sized satellites could detect nuclear bombs in space, and a World Cup match drove Google Search traffic to a new record.

Researchers at Anthropic have developed a new tool called the Jacobian lens, or J-lens, to peer into the inner workings of their large language model, Claude. The tool revealed a hidden region, named J-space, where the model processes words related to potential responses it may never actually output. This discovery offers an unprecedented look at how AI models reason before finalizing their answers, revealing both routine and unsettling patterns in their decision-making.

Meanwhile, OpenAI has launched its long-awaited “super app,” ChatGPT Work, which combines its chatbot, coding assistant, and new models into a single platform. Designed to handle tasks both for and alongside users, the app debuted alongside OpenAI’s GPT 5.6 models. The company is also developing a fully automated researcher, signaling its push toward more autonomous AI agents.

In other developments, humanoid robots have performed teleoperated surgery on living animals for the first time, removing gallbladders from pigs. SK Hynix, a South Korean chip giant, raised $26.5 billion in the largest U.S. listing by a foreign company, driven by surging demand for AI data centers. Tencent is leading a deal to unwind Meta’s $2 billion acquisition of Chinese AI startup Manus, following Beijing’s order to divest.

Elsewhere, scientists revived light responses in human retinas 10 hours after death, advancing eye transplant research. Meta has introduced paid tiers for its Muse Spark AI tool and plans to produce its own AI chip by September. OpenAI and Google reportedly sold AI models to blacklisted Chinese groups via Singapore-based subsidiaries. These stories highlight the rapid, sometimes controversial, evolution of AI technology.

A third-generation California dairy farmer, Anthony Agueda, is turning to an unlikely ally in the fight against agricultural pollution: red earthworms. By raking through a bed of dark, wet wood chips, he reveals dozens of the creatures, with hundreds of thousands more hidden beneath the surface. These worms, combined with microbes, form a system called vermifiltration that cleans manure wastewater, potentially slashing emissions of methane and nitrous oxide while reducing water contamination. This approach is gaining traction among farmers, companies, and scientists as the livestock industry faces mounting pressure to address its environmental footprint.

Meanwhile, the field of solar geoengineering is shifting from theoretical computer models to practical engineering challenges. Researchers are now developing aircraft, materials, and other systems needed to deliberately intervene in the climate system to counteract global warming. However, as they dive into the details, they are discovering that even an early deployment would require substantial new infrastructure, significant time, and major investment.

These developments highlight the growing urgency to tackle environmental issues from multiple angles. From humble earthworms reshaping sustainable farming to the complex realities of geoengineering, innovators are pushing boundaries. For more on these stories, explore the latest insights from MIT Technology Review AI.

Sam Altman’s proposal to give the U.S. government a 5% stake in OpenAI has resurfaced, potentially worth about $320 per American household at current valuation. The idea aims to address concerns that AI companies profit from human labor without compensating creators, while also providing a safety net against job market disruption. However, details remain vague, and experts suggest the offer may serve more as a political narrative than a workable policy, according to MIT Technology Review AI.

Meanwhile, a leaked Treasury report warns that the AI market resembles the dotcom bubble, contradicting the administration’s public optimism. Fears of overinflation are growing, with AI profits masking broader risks in earnings reports. This caution comes as Samsung reports a 1,800% profit surge from AI chip sales, though shares slumped on worries the boom may stall.

In other AI news, Illinois’ governor signed the nation’s strongest frontier AI law to protect citizens from risks, while U.S. lawmakers clash over federal rules. A hidden tracker in Anthropic’s Claude Code was exposed and removed after secretly monitoring users in China, raising surveillance concerns. Additionally, a controversial AI “actor” named Tilly Norwood is set to star in its first feature film, sparking backlash from a major actors union.

South Korean chipmaker SK Hynix has made history by raising $26.5 billion in the largest foreign initial public offering ever on U.S. markets, surpassing Alibaba's 2014 record. The company sold nearly 178 million American depositary shares at $149 each, allowing U.S. investors to buy in at roughly one-tenth the price of a full share in Seoul. Trading began Friday on the Nasdaq under a temporary ticker, with shares surging 14% above the IPO price amid demand that reportedly exceeded supply by seven times.