The new sites will boost Stargate's planned capacity to nearly 7 gigawatts--about equal to the output of seven large nuclear reactors. An aerial view shows construction underway on a Project Stargate AI infrastructure site in Abilene, Texas on April 23, 2025. OpenAI is planning to build five new data centers in the United States as part of the Stargate initiative, the company announced on Tuesday. The sites, which are being developed in partnership with Oracle and SoftBank, bring Stargate's current planned capacity to nearly 7 gigawatts--roughly the same amount of power as seven large-scale nuclear reactors . "AI is different from the internet in a lot of ways, but one of them is just how much infrastructure it takes," OpenAI CEO Sam Altman said during a press briefing in Abilene, Texas on Tuesday.

Watch a subscriber-only conversation with Sneha Goenka, who designed the computations behind the world's fastest whole-genome sequencing method. Every year, MIT Technology Review selects one individual whose work we admire to recognize as Innovator of the Year. For 2025, we chose Sneha Goenka, who designed the computations behind the world's fastest whole-genome sequencing method . Thanks to her work, physicians can now sequence a patient's genome and diagnose a genetic condition in less than eight hours--an achievement that could transform medical care. Exclusive: A record-breaking baby has been born from an embryo that's over 30 years old Jessica Hamzelou Therapists are secretly using ChatGPT. Exclusive: A record-breaking baby has been born from an embryo that's over 30 years old The embryos were created in 1994, while the expectant father was still a toddler, and donated via a Christian "embryo adoption" agency.

Membership in ACM includes a subscription to Communications of the ACM (CACM), the computing industry's most trusted source for staying connected to the world of advanced computing. Can AI Expand the Human Mind? LLMs could represent a new layer of human cognition, which researchers call'System 0.' Giuseppe Riva first started to think about the role that artificial intelligence (AI) can play in human cognition when he and a colleague were trying to find someplace to have dinner in Los Angeles. Both pulled out their phones and started perusing Google Maps for suggestions of nearby restaurants. Riva quickly noticed that the list of possibilities on his phone was very different from what his companion was seeing.

Membership in ACM includes a subscription to Communications of the ACM (CACM), the computing industry's most trusted source for staying connected to the world of advanced computing. Why we should be intentional about data scaling. Large language models (LLMs) have revolutionized the AI landscape, demonstrating remarkable capabilities across a wide range of tasks. Each new model seemingly reinforces the notion that modern transformer-based AI can conquer any challenge if armed with sufficient compute and data. However, while scaling has accelerated certain applications, such as robotics, it has yet to show significant impact in others, such as identifying misinformation.

Membership in ACM includes a subscription to Communications of the ACM (CACM), the computing industry's most trusted source for staying connected to the world of advanced computing. Addressing the Arab world's unique challenges against misinformation and disinformation requires efforts at technical, institutional, and social levels. Misinformation and disinformation are global risks. However, the Arab region is particularly vulnerable due to its geopolitical instabilities, linguistic diversity, and other cultural nuances. Misinformation includes false or misleading content, such as rumors, satire taken as fact, or conspiracy theories, while disinformation is the intentional and targeted spread of such content to deceive or manipulate specific audiences. To limit the spread and influence of misinformation, it is essential to advance research on technological methods for early detection, tracking, and mitigation, while also strengthening media literacy and promoting active citizen participation.

Venture capitalist Alexa von Tobel is ready to bet on quantum computing--starting with hardware. Alexa von Tobel is fully aware that her big bet on quantum computing may never pay off. "The risk of being too early is a real risk," she says. She's speaking to me via Zoom from the New York City office of Inspired Capital, the early-stage venture capital firm she runs with former US commerce secretary Penny Pritzker. In addition to personally investing in blue-chip brands like Uber and Airtable, von Tobel has backed a number of AI startups through Inspired Capital, including BrightAI (a platform that monitors critical infrastructure) and PreemptiveAI (a startup building a foundation model to map human physiology and predict health outcomes).

When you purchase through links in our articles, we may earn a small commission. Sick of AI in your Windows 11 PC? Here's how to get rid of it Microsoft wants to put AI everywhere on your PC, but you can take back control. Microsoft is flooding every inch of Windows 11 with AI features and if you're like me, you aren't thrilled about it. I'd rather choose the AI tools I want to use instead of Microsoft shoving them into my PC unprompted. The good news is, you can turn off the AI features in Windows 11.

Large language models (LLMs) can now access a wide range of external tools, thanks to the Model Context Protocol (MCP). This greatly expands their abilities as various agents. However, LLMs rely entirely on the text descriptions of tools to decide which ones to use--a process that is surprisingly fragile. In this work, we expose a vulnerability in prevalent tool/function-calling protocols by investigating a series of edits to tool descriptions, some of which can drastically increase a tool's usage from LLMs when competing with alternatives. Through controlled experiments, we show that tools with properly edited descriptions receive over 10 times more usage from GPT-4.1 and Qwen2.5-7B than tools with original descriptions. We further evaluate how various edits to tool descriptions perform when competing directly with one another and how these trends generalize or differ across a broader set of 17 different models. These phenomena, while giving developers a powerful way to promote their tools, underscore the need for a more reliable foundation for agentic LLMs to select and utilize tools and resources. Our code is publicly available at https://github.com/kazemf78/llm-unreliable-tool-preferences.

Language models obtain extensive capabilities through pre-training. However, the pre-training process remains a black box. In this work, we track linear interpretable feature evolution across pre-training snapshots using a sparse dictionary learning method called crosscoders. We find that most features begin to form around a specific point, while more complex patterns emerge in later training stages. Feature attribution analyses reveal causal connections between feature evolution and downstream performance. Our feature-level observations are highly consistent with previous findings on Transformer's two-stage learning process, which we term a statistical learning phase and a feature learning phase. Our work opens up the possibility to track fine-grained representation progress during language model learning dynamics.

Generative models for counterfactual outcomes face two key sources of bias. Confounding bias arises when approaches fail to account for systematic differences between those who receive the intervention and those who do not. Misspecification bias arises when methods attempt to address confounding through estimation of an auxiliary model, but specify it incorrectly. We introduce DoubleGen, a doubly robust framework that modifies generative modeling training objectives to mitigate these biases. The new objectives rely on two auxiliaries -- a propensity and outcome model -- and successfully address confounding bias even if only one of them is correct. We provide finite-sample guarantees for this robustness property. We further establish conditions under which DoubleGen achieves oracle optimality -- matching the convergence rates standard approaches would enjoy if interventional data were available -- and minimax rate optimality. We illustrate DoubleGen with three examples: diffusion models, flow matching, and autoregressive language models.