AI Daily Briefing — 2026-06-28

A blistering heatwave sweeping Western Europe, with London hitting a record 36.1°C this week, is doing more than wilting crops and straining power grids—it's also frying people's brains. Researchers are increasingly concerned about how soaring temperatures impact mental function and behavior. Studies show that as the mercury climbs, people become more irritable and aggressive, while firefighters struggle to concentrate after exposure to extreme heat. The effects are especially dangerous for children and those with mental health conditions, according to experts cited by MIT Technology Review AI.

Animal studies suggest that excessive heat can disrupt chemical signaling in the brain, but scientists are still working to understand the exact mechanisms at play. The findings underscore a growing public health concern as climate change makes heatwaves more frequent and intense. Beyond the immediate physical dangers, the cognitive toll of extreme heat could have far-reaching consequences for safety, productivity, and social stability.

In other tech news, the Trump administration has imposed unprecedented restrictions on OpenAI, requiring government vetting of the first users of its upcoming GPT-5 model before a wider release. It marks the first time a US firm has been ordered to limit an AI model's launch. Meanwhile, Apple and Xbox have raised prices on products like MacBooks and iPads by over 20%, blaming surging costs for AI-driven memory and storage chips—a shortage dubbed 'RAMaggedon.'

OpenAI announced Friday it is limiting the release of its latest GPT-5.6 family of AI models to a small group of trusted partners, following a request from the Trump administration. The lineup includes Sol, the flagship model; Terra, a balanced everyday option; and Luna, a faster, cheaper variant. The company stated that the preview is restricted to partners whose participation has been disclosed to the government, marking a significant shift in how advanced AI systems are deployed.

The administration’s move comes amid heightened pressure on AI developers to limit access to cutting-edge models. Earlier, Anthropic was forced to pull its Fable 5 model after being ordered to block foreign nationals from using it. Dean Ball, a former White House AI adviser and incoming OpenAI employee, argued that a recent executive order requiring voluntary pre-release government review has created an informal licensing system. Without clear safety standards, he warned, such delays could hinder U.S. competitiveness and threaten billions in AI infrastructure investments.

OpenAI emphasized that it complied with the request reluctantly, stating in a blog post that this process should not become the norm. “It keeps the best tools from users, developers, enterprises, and global partners who need them,” the company wrote. OpenAI described the limited rollout as a short-term step toward broader availability in the coming weeks, as it works with the administration to develop a new executive order on cybersecurity and a repeatable framework for future releases.

The GPT-5.6 Sol model boasts enhanced agentic capabilities in coding, biology, and cybersecurity, with a new “ultra” reasoning mode using coordinated subagents for complex tasks. OpenAI claims Sol outperforms Anthropic’s Claude Mythos 5 on coding benchmarks while using fewer output tokens. The company also highlighted Sol’s robust security stack, designed to resist jailbreaking and prioritize defensive cybersecurity over offensive exploits.

Nvidia's long reign over the AI chip market is facing its most significant challenge yet, as a wave of major companies race to develop their own silicon. OpenAI recently unveiled plans for Jalapeño, a custom inference chip built in partnership with Broadcom, joining a growing list of firms including Google, Apple, and SpaceX. These moves signal a strategic shift away from total reliance on a single supplier, aiming instead for greater control and hardware optimized for specific tasks.

The trend is less about abandoning Nvidia entirely and more about creating a hedge against supply chain risks. Custom silicon allows companies to tune performance to their exact needs, much like Apple did when it transitioned away from Intel processors. This approach promises efficiency gains and cost savings, particularly for massive-scale AI operations where even small optimizations can yield substantial benefits.

On a recent episode of TechCrunch's Equity podcast, hosts Kirsten Korosec, Anthony Ha, and Sean O'Kane explored how this custom chip movement could reshape the industry. They discussed the implications for Nvidia's market position and the broader competitive landscape, noting that the shift could accelerate innovation across the AI hardware ecosystem. As more players invest in bespoke solutions, the era of one-size-fits-all AI chips may be drawing to a close.

Europe is grappling with a record-breaking heat wave that is pushing its power grid to the brink. As residents crank up fans and air conditioners to stay cool, demand for electricity has surged. However, many power plants are offline due to planned maintenance, originally scheduled for spring and summer when demand was historically lower. This seasonal shift, driven by climate change, is creating a dangerous mismatch between supply and need, according to Casey Crownhart of MIT Technology Review AI. Utilities now face the urgent challenge of adapting to more frequent and intense heat waves while rapidly expanding capacity.

Meanwhile, IBM has announced a breakthrough in chip technology that could extend Moore's Law for another decade. The company built a prototype chip with roughly 100 billion transistors on an area the size of a fingernail—double the density of its 2021 design. As transistors approach physical limits where shrinking them further degrades performance, IBM's solution involves stacking components vertically, much like building up in crowded cities. This approach promises faster, more energy-efficient computers in the years ahead.

The dual stories highlight how technology must evolve to address climate pressures and computing limits. As Europe scrambles to keep the lights on, IBM's innovation offers a glimpse of a future where chips continue to advance despite physical barriers.

MIT Technology Review AI introduces its new Engineering issue, a collection of stories that highlight humanity’s drive to solve problems through ingenuity. From tunneling beneath the seafloor to nanoscale chipmaking advances, the issue explores efforts at every scale—including a controversial plan to mimic volcanic cooling to lower Earth’s temperature. The message is clear: while we can’t fix everything, ambitious engineering can deliver real progress.

In other news, Stripe, Anthropic, and OpenAI are backing a $500 million nonprofit aimed at preventing respiratory infections like the common cold and flu. The venture, also supported by Bill Gates, will explore whether modern technology can eliminate these viruses entirely. Meanwhile, astronomers recount the dramatic hunt for asteroid 2024 YR4, which briefly posed the highest impact risk ever recorded before being ruled safe.

China has reclaimed the title of world’s fastest supercomputer with Shenzhen’s LineShine, surpassing California’s El Capitan. Separately, Anthropic’s AI model reportedly found flaws in classified US systems, sparking security concerns and a suspension. These stories and more are featured in the latest issue, available now.