AI This Week

Microsoft is moving away from exclusive reliance on OpenAI by integrating Anthropic’s Claude Sonnet 4 AI into Office 365 applications like Word, Excel, PowerPoint, and Outlook. This decision reflects growing tension between Microsoft and OpenAI, as both companies pursue independent initiatives. Microsoft’s leaders reportedly believe Anthropic's models excel in specific capabilities, such as designing visually appealing presentations, and see value in diversifying their AI portfolio. Additionally, Microsoft is advancing its own in-house AI systems while negotiating new terms with OpenAI, whose restructuring plans remain in progress. On the other side, OpenAI is reducing dependence on Microsoft’s infrastructure by developing its own AI chips. This evolving competition marks a shakeup in the AI market.

Alibaba Group has introduced its most advanced AI model, Qwen-3-Max-Preview, featuring over 1 trillion parameters. Released via its cloud platform and OpenRouter marketplace, this model joins the elite ranks of AI systems developed by OpenAI and Google DeepMind. A member of the company’s Qwen3 series, Qwen-3-Max-Preview is designed exclusively for text-based tasks and outperformed earlier iterations in internal evaluations. It reportedly surpassed competitors like MoonShot AI’s Kimi K2 and Anthropic’s Claude Opus 4 in performance benchmarks, though Alibaba has not provided an official technical report. AI parameters, which determine a model’s learning capacity, continue to be a key measure of innovation in the field.

French AI firm Mistral has solidified its position as Europe’s leading AI contender with a staggering €17 billion investment round. The funding, led by Dutch chip manufacturer ASML, granted the sponsor an 11% stake and a seat on Mistral’s strategic committee. The company’s valuation doubled to €117 billion, showcasing rapid growth yet highlighting the gap compared to U.S. rival Anthropic’s €183 billion mark. Mistral stands apart through its commitment to open-source AI models, allowing developers worldwide to adapt and improve them. Strategic collaborations, including NVIDIA and MGX, reflect Mistral’s ambitions for global AI influence. By partnering with ASML, Mistral aims to advance innovation across both AI models and semiconductor technology

Warner Bros. has filed a lawsuit against Midjourney, claiming the AI platform violates copyrights by generating unauthorized images and videos of its iconic characters, including Superman, Batman, and Bugs Bunny. The studio alleges that Midjourney deliberately removed safeguards that could have prevented such infringements. This case follows a similar template to earlier lawsuits filed by Disney and Universal, suggesting a united front by major studios against AI misuse. Midjourney, on the other hand, argues its practices fall under “fair use” and accuses the studios of double standards. The platform’s expansion into video services and streaming channels has heightened tensions, with studios viewing this as a competitive threat. Warner Bros. seeks damages and an injunction.

Biotech startup Enveda is using artificial intelligence to search thousands of plants for potential drug breakthroughs. The Colorado-based company recently achieved unicorn status with a $150 million funding round, valuing it at over $1 billion. Enveda's approach combines traditional plant-based medicine knowledge with AI to identify promising molecules for drug development. The firm has already begun its first clinical trial and plans to expand its pipeline. This innovative method aims to accelerate the discovery of new medicines from natural sources.

Inside Higher Ed's Student Voice survey reveals widespread adoption of generative AI among college students, with 85% using it for coursework. Students primarily utilize AI for brainstorming, tutoring, and studying. While many find AI helpful, some worry about its impact on critical thinking. Nearly all students want clear institutional policies on AI use, preferring guidance over restrictions. Despite AI's prevalence, most students still view college as valuable or increasingly important in the AI era. The survey included responses from 1,047 students across 166 institutions.

PC Gamer reports on Nvidia's efforts to improve AI reasoning capabilities. The company is using human tutors to teach its Cosmos Reason AI model basic knowledge about the physical world through multiple-choice questions based on video data. This approach aims to develop AI that can safely interact with real-world environments, particularly for applications in robotics and autonomous vehicles. Nvidia's data factory team, comprising experts from various fields, is compiling and analyzing data to enhance the AI's understanding of everyday scenarios.

Scientists at UC Riverside have introduced a groundbreaking method to erase private and copyrighted data from AI models without requiring access to the original training datasets. Their "source-free certified unlearning" approach substitutes original data with a surrogate dataset and injects calibrated random noise to alter model parameters, ensuring selected information is irretrievable. The technique maintains the functionality of the models while reducing the costly and energy-intensive need for complete retraining. This innovation addresses growing legal and ethical concerns, such as compliance with privacy regulations and protection of copyrighted material used to train AI systems like GPT. Researchers demonstrated effectiveness using both synthetic and real-world datasets, offering strong privacy assurances.

The United Nations General Assembly has approved two groundbreaking initiatives aimed at shaping global cooperation on artificial intelligence governance. These include the establishment of the Independent International Scientific Panel on AI and the Global Dialogue on AI Governance. The Scientific Panel is designed to bridge cutting-edge AI research and policy-making while delivering independent assessments to inform critical decisions. Meanwhile, the Global Dialogue will unite nations and stakeholders to discuss pressing AI challenges. Both mechanisms build upon the Global Digital Compact adopted in September as part of the Pact for the Future. The UN Secretary-General, António Guterres, will soon launch nominations for the Scientific Panel, which will begin publishing annual reports in 2026.

First Lady Melania Trump has launched a nationwide artificial intelligence challenge for K-12 students. The initiative encourages young participants to create projects using AI to address community challenges. In her video announcement, Trump emphasized the importance of early exposure to AI, inviting educators to design creative teaching approaches involving the technology. Trump referenced her own experience with AI, including an AI-narrated version of her memoir "Melania," as an example of the potential of the technology. President Donald Trump backed the initiative during a Cabinet meeting, directing attention to the AI.gov website. The site stresses the importance of responsible AI training for students to foster innovation and economic growth. The challenge highlights growing efforts to prepare the next generation for an AI-driven future.