FutureGenNews

Jensen Huang, Alice Huang and the China Spy Allegations Explained

This prediction from 1964 about our AI future is 100% accurate.

A 15-year-old student in Japan was arrested after allegedly using a ChatGPT-assisted program to cancel more than 46,000 Bandai Channel anime subscriptions. Police say he found a system vulnerability, sent false data to the company’s servers, and disrupted the service for weeks. This shows the darker side of AI-powered coding: even young users can now build tools that expose weak security systems. AI is becoming more powerful, but cybersecurity needs to catch up faster. Source: Asahi/Kyodo

Meta reportedly paid hundreds of contractors to pose as teens and test rival AI chatbots, per WIRED. Using fake teen accounts, they asked ChatGPT, Gemini and CharacterAI thousands of prompts on self-harm, eating disorders, sex and drugs. AI safety testing is getting intense.

This 6-minute answer from the Godmother of AI, Fei-Fei Li, might change how you think about the future of intelligence.

Google is reportedly restricting Meta's use of Gemini due to capacity shortages, per the Financial Times. The impact: Meta's internal projects for customer support and content moderation are now facing delays. This is a fascinating moment in the AI wars. We're no longer just competing on model quality or pricing. Now we're competing on compute availability. Google can literally throttle a rival by limiting API access when capacity is tight. Your point about token efficiency is spot-on. In a world where compute is the bottleneck, the companies that can do more with fewer tokens—or that can build their own infrastructure—will have a massive advantage. The next moat in AI may not be better models. It may be better efficiency and independent compute. #Google #Meta #AI #TechNews #Gemini

Google just added a new feature to Gemini in Chrome: it can now see what you're looking at on a webpage. Highlight text or click an image, and Gemini automatically uses it in your question. No more copy-paste. No more screenshots. You can ask Gemini to explain a chart, summarize an article, compare products, identify an image, or answer questions about anything on the page in real time. This is a subtle but significant shift. The browser is becoming the interface between you and AI. Instead of switching between tabs and tools, you stay on the page and let Gemini see what you see. The next productivity jump may not come from better AI models. It may come from AI that is so integrated into your workflow that you barely notice you're using it.

Helios may be one of the most important quantum computing milestones yet. Quantinuum’s new trapped-ion quantum computer, Helios, uses 98 barium-ion qubits suspended and controlled with extreme precision. The real breakthrough is not just the qubit count, but the accuracy: very low error rates, all-to-all connectivity, and performance that pushes beyond easy classical simulation. What makes this important is that quantum computing is not a race for more qubits alone. It is a race for better qubits. Helios combines scale, precision, programmability, and a QCCD “quantum railway” architecture where ions can be moved between memory and operation zones. This does not mean useful quantum computers are suddenly here. But it does show the field is moving from impressive lab demos toward real computing systems. The quantum revolution probably will not arrive in one giant leap. It will be engineered step by step, atom by atom.

OpenAI just unveiled Jalapeño, its first custom AI inference chip built with Broadcom. This is bigger than just hardware. Inference is what powers every ChatGPT response, API call, and AI agent task at scale. If OpenAI can make AI faster, cheaper, and more energy-efficient, it controls more of the full AI stack. The AI race is no longer just about models. It’s becoming a compute war.

NVIDIA says its new AI cooling system can run liquid coolant at up to 45°C / 113°F — hotter than a hot tub. Why it matters: future AI data centers could use closed-loop liquid cooling, dry coolers, and almost no facility cooling water in the right climates. AI is not just a chip race anymore. It is an infrastructure race. Source: NVIDIA Blog #NVIDIA #AI #DataCenters #LiquidCooling

Meta is giving free Ray-Ban Meta AI glasses to every legally blind veteran in the United States. More than 130,000 veterans are eligible. The program was inspired by U.S. Army veteran Don Overton, who used the glasses to regain more independence in daily life. With AI and voice commands, the glasses can read documents and signs, identify nearby objects, answer questions about surroundings, handle calls, and help with everyday tasks. Each recipient will also get training and support through webinars, in-person events, and a guide built specifically for blind and low-vision veterans. This is one of the strongest real-world use cases for AI wearables so far. Not just notifications on your face. A tool that helps people navigate the world more independently.

SpaceX is acquiring Cursor in a massive $60 billion all-stock deal. If completed, Cursor will become a wholly owned SpaceX subsidiary, with shareholders receiving SpaceX Class A shares based on SpaceX’s 7-day average share price before closing. The deal is expected to close in Q3 2026, pending regulatory approvals and customary conditions. This would be one of the largest AI acquisitions ever. Cursor is already one of the fastest-growing AI coding platforms in the world. Bringing it under SpaceX would dramatically expand SpaceX’s AI footprint beyond rockets, satellites, and infrastructure. The bigger signal: AI coding tools are becoming strategic assets for frontier engineering companies. SpaceX may not just want better software. It may want an AI-native engineering stack for everything it builds.

Europe’s next major defense race is taking shape around AI-powered “wingman” aircraft. These are unmanned combat jets designed to fly alongside fighter pilots, carrying extra sensors, jammers, weapons, and autonomous decision-support systems. At the Berlin airshow, Airbus, Boeing, Helsing, and General Atomics all pushed their versions as Europe rearms and tries to reduce reliance on U.S. defense technology. Boeing says its MQ-28 Ghost Bat could enter German Luftwaffe service by 2029, while Airbus’ Ravenstorm is expected in the 2030s. The key lesson from Ukraine is clear: future air wars won’t just be about jets. They’ll be about drones, electronic warfare, data networks, and AI agents operating around human pilots. Europe is no longer just buying aircraft. It is trying to build the operating system for AI warfare. Source: Reuters

Researchers in Singapore have developed a seed-sized surgical robot that can switch between five medical tools in under one second. The robot is just 4.4 mm long and is controlled wirelessly using magnetic fields. What makes it wild is the 5-in-1 design: It can navigate inside the body, cut tissue, deliver drugs, collect biopsy samples, and generate localized heat for potential cancer treatments. In lab tests, it successfully moved across soft biological surfaces while cutting tissue, delivering particles, collecting samples, and producing heat. This is where surgery is heading. Less invasive procedures. More targeted treatments. And tiny robotic systems that can perform multiple tasks inside the body without needing large surgical tools. The future operating room may look less like open surgery and more like remote-controlled micro-robotics. Source: NTU Singapore / Advanced Materials

OpenAI is reportedly considering drastic price cuts to win customers away from Anthropic, according to the WSJ. That tells you how intense the AI race has become. For the last two years, the competition was mostly about who had the best model. Now it is becoming a pricing war. If OpenAI starts cutting aggressively, Anthropic will be forced to respond, and enterprise AI buyers could suddenly get much more leverage. The cost of intelligence keeps collapsing. And the companies spending billions to build it may now have to fight on margins too. The next phase of the AI war may not be benchmarks. It may be discounts.