AI Firm Fundamental Launches NEXUS to Streamline Tabular Data Analysis
San Francisco-based AI firm Fundamental launched NEXUS, a native foundation model designed to bypass manual ETL (Extract, Transform, Load) processes for tabular data, according to VentureBeat on February 5, 2026. The new system aims to address the "blind spot" in the deep learning revolution, where structured, relational data in spreadsheets and databases has been treated as just another file format.
NEXUS intends to automate the labor-intensive data science process of manual feature engineering and classic machine learning algorithms that predate modern deep learning. Fundamental, co-founded by DeepMind alumni, seeks to provide enterprises with a more efficient way to forecast business outcomes using AI.
Gen Z Users Shift Away From TikTok USA
Meanwhile, Gen Z users in the United States are reportedly shifting away from TikTok following its restructuring under a new joint venture with Oracle, Fortune reported. The shift, framed by some as a "quiet revolt," involves users downloading an alternative app created by a former Oracle employee.
In January, TikTok's U.S. operations were formally separated from its global business, with Oracle taking on responsibility for American user data and a U.S.-run version of TikTok's recommendation algorithm. This move, intended to address national security concerns, was perceived by many young users as the app becoming "an instrument of corporate and political power," according to Fortune. Creators have reportedly voiced their discontent on TikTok itself.
Concerns Raised Over AI Model Evaluation and Capabilities
The AI community continues to monitor the development of large language models (LLMs) closely, with particular attention paid to a graph maintained by the AI research nonprofit METR (Model Evaluation & Threat Research), according to MIT Technology Review. This graph, released in March of last year, tracks the development of certain AI capabilities and suggests they are developing at an exponential rate. Recent model releases, such as Claude Opus 4.5, have reportedly outperformed this trend.
LinkedIn Probes for Chrome Extensions
In other tech news, GitHub users have discovered that LinkedIn silently probes for 2,953 Chrome extensions on every page load, according to a repository documenting the findings. The repository provides a list of the extensions, their names, and links to the Chrome Web Store. The discovery raises privacy concerns about the extent of LinkedIn's data collection practices.
Nature Publishes Correction Regarding Colibactin-Driven Colon Cancer Research
Nature News issued a publisher correction to an article published on November 6, 2024, regarding colibactin-driven colon cancer. The correction addresses errors in labeling figures within the original article, specifically in Figs. 2 and 3. The corrected labels involve the terms "ΔfimH" and "ΔfmlH" in relation to experimental models of disease. The publisher noted that the errors have been corrected in the online version of the article.
Discussion
AI Experts & Community
Be the first to comment