Two Chinese nationals, Chuan Geng and Shiwei Yang, have been accused of illegally shipping tens of millions of dollars' worth of Nvidia AI chips to China without required licenses from the US Commerce Department. The chips, including advanced Nvidia H100s, were exported by ALX Solutions, a company founded in 2022 by Geng and Yang, between October 2022 and July 2025. The company received payments from Chinese companies, including a $1 million payment in January 2024, and shipped the chips through Singapore and Malaysia to China, often used as transshipment points for illegal goods.
NVIDIA's Physical AI Dataset is a 15-terabyte collection of carefully structured trajectories, designed to reflect operational complexity. This dataset is crucial for robot operating systems to operate with precision in complex real-world environments. The cost of errors in physical systems can be high, as seen in self-driving robots misjudging pallet dimensions or surgical-assistant robots mistaking flexible instruments for rigid tissue. To address this, Physical AI leaders prioritize curated, domain-specific datasets over brute-force scale. This requires defining physical fidelity metrics, curating and annotating with domain expertise, and iterating with closed-loop feedback. Companies that invest in high-quality datasets will have an edge in critical infrastructure, such as logistics centers and hospitals, where reliability and user trust are paramount.
OpenAI has released two open-weight language models that excel in advanced reasoning, optimized to run on laptops with performance levels similar to its smaller proprietary reasoning models. The models were trained on a text-only dataset focusing on science, math and coding knowledge, and are distinct from open-source models.
A recent report by IESE's Education for Jobs Initiative highlights significant gaps in innovation, execution, and analytical skills among companies in Spain, with 92% reporting major difficulties in these areas. The survey of 81 leading companies across 18 sectors found that AI adoption is slow, with only 7% completing more than half of their implementation process. Critical capabilities needed for successful AI integration include data analytics, machine learning, and AI system development. However, ethical values around AI use are also considered fundamental competencies. The report notes that remote work has reshaped skills required in professionals, with learning agility, flexibility, and autonomy topping the list. Leaders must now possess clear communication, empathy, and motivation to lead virtual teams effectively. Demographic shifts pose a challenge for companies, particularly in Spain where over 30% of the population is expected to be aged 65 or over by 2050.
Elon Musk's AI startup, xAI, plans to release its chatbot Grok 2 as open-source software next week. This move is part of a growing trend in the tech industry where companies are making their AI technologies more accessible to developers and researchers. The open-sourcing of Grok 2 aims to accelerate innovation and collaboration in the field of conversational AI, allowing experts from around the world to build upon and improve the technology.
Appdome uses AI to generate defense plugins for mobile apps, providing security features such as anti-fraud, bot defense, and geo compliance. The company's ThreatScope Mobile XTM platform continuously calculates a Mobile Risk Index for businesses and applications, ranking and preempting attacks in real-time. Appdome's Threat Resolution Center utilizes Agentic-AI to provide customer support teams with tools to resolve threats quickly. The platform also tracks all builds, changes, and defense configurations for easy audit of the mobile defense lifecycle, holding several patents including U.S. Patents 9,934,017 B2 and 11,294,663 B2.
The Federal Aviation Administration (FAA) has proposed rules to normalize Beyond Visual Line of Sight (BVLOS) drone operations, which could unlock transformative scaled drone applications across key sectors such as agriculture, energy, and public safety. Safe Pro Group Inc., a leader in artificial intelligence-powered defense and security solutions, expects these new rules to expand the use of its AI-powered computer vision and object detection technologies, accelerating the deployment of AI-powered drone systems for government, public, and private customers. The proposed rules would enable greater operational flexibility for operators and drone manufacturers, with detailed requirements for operations, aircraft manufacturing, and safety protocols. Safe Pro's Airborne Response unit is one of Florida's top providers of drone-based critical infrastructure inspections, trusted by utilities, telecoms, and insurers for real-time insights.
Google has committed $1 billion over three years to provide AI training and tools to US higher education institutions and nonprofits, with over 100 universities already signed up. Participating schools may receive cash funding and resources such as cloud computing credits for AI training. The initiative aims to expand to every accredited nonprofit college in the US and other countries, with Google giving advanced AI tools like the Gemini chatbot to college students for free.
Elon Musk's AI startup xAI plans to release its Grok 2 chatbot as open-source software next week, aiming to accelerate innovation and collaboration in conversational AI by making the technology more accessible to developers and researchers worldwide.