Reid Hoffman Thinks Doctors Should Ask AI for a Second Opinion

The hidden cost of Google's AI defaults and the illusion of choice

Many people are hoping—nay, praying—that the potential AI bubble will burst soon. But to hear Google tell it, generative AI is the future, and the company’s products have to change to keep up with the technical reality. As a result, Gemini is seeping into every nook and cranny of the Google ecosystem. Generative AI feeds on data, and Google has a lot of your data in products like Gmail and Drive. What does that mean for your privacy, and what happens if you don’t want Gemini peeking over your shoulder? Well, it’s kind of a mess. The amount of data Gemini retains depends on how you access the AI, and opting out of data collection can mean running straight into so-called “dark patterns,” UI elements that work against the user’s interest. This is the future? Google doesn’t train AI with…

Researchers try to cut the genetic code from 20 to 19 amino acids

The genetic code is central to life. With minor variations, everything uses the same sets of three DNA bases to encode the same 20 amino acids. We have discovered no major exceptions to this, leading researchers to conclude that this code probably dated back to the last common ancestor of all life on Earth. But there has been a lot of informed speculation about how that genetic code initially evolved. Most hypotheses suggest that earlier forms of life had partial genetic codes and used fewer than 20 amino acids. To test these hypotheses, a team from Columbia and Harvard decided to see if they could get rid of one of the 20 currently in use. And, as a first attempt, they engineered a portion of the ribosome that worked without using an otherwise essential amino acid: isoleucine. Changing the code…

Meta cuts contractors who reported seeing Ray-Ban Meta users have sex

In February, numerous workers from a company that Meta contracted to perform data annotation for Ray-Ban Meta reported viewing sensitive, embarrassing, and seemingly private footage recorded by the smart glasses. About two months later, Meta ended its contract with the firm. According to a BBC report today, “less than two months” after a report from Swedish newspapers Svenska Dagbladet and Göteborgs-Posten and Kenya-based freelance journalist Naipanoi Lepapa came out featuring Sama workers complaining about watching explicit footage shot from Ray-Ban Metas, “Meta ended its contract with Sama.” Sama is a Kenya-headquartered firm that Meta contracted to perform data annotation work, including working with video, image, and speech annotation for Meta’s AI systems for Ray-Ban Metas. Sama claims that Meta’s cancellation of the contract affected 1,108 workers. A Meta spokesperson told BBC that Meta “decided to end our work with Sama…

Elon Musk's 7 biggest stumbles on the stand at OpenAI trial

Elon Musk seems tired and cranky. On Thursday, he took the stand for the third day in a four-week trial stemming from his lawsuit alleging that OpenAI abandoned its mission and should be blocked from taking the company public later this year. If Musk plays his cards right, Sam Altman could be ousted and OpenAI would remain a nonprofit forever. But Musk stumbled at least seven times in ways that possibly put his chances at winning in jeopardy. Most notable, 1) OpenAI’s lawyer managed to get him to make several concessions over his own lawyer’s objections. 2) He also lost a fight to keep xAI’s safety record off the table, calling his reputation as a supposed AI savior defending OpenAI’s mission into question. 3) He repeatedly appeared dishonest, as OpenAI’s lawyer showed documents contradicting his testimony. And he twice appeared…

Sam Altman is “the face of evil” for not reporting school shooter, says lawyer

OpenAI could have prevented one of the deadliest mass shootings in Canada’s history, a string of seven lawsuits filed Wednesday in a California court alleged. Ultimately, the AI company overruled recommendations from its internal safety team. More than eight months prior to the school shooting, trained experts had flagged a ChatGPT account later linked to the shooter as posing a credible threat of gun violence in the real world. In those cases, OpenAI is expected to notify police—which, in this case, already had a file on the shooter and had proactively removed guns from their home previously—but that’s not what happened. Apparently, OpenAI decided that the user’s privacy and the potential stress of an encounter with cops outweighed the risks of violence, whistleblowers told The Wall Street Journal. Leaders rejected the safety team’s urgings and declined to report the user…

Drone strikes on data centers spook Big Tech, halting Middle East projects

A data center developer has paused all Middle East project investments after one of its facilities was damaged by an Iranian missile or drone attack. The decision comes as the Iran war is forcing Silicon Valley investors and tech companies to rethink a trillion-dollar plan to build more AI and cloud data centers in Gulf countries. The damaged data center is owned by Pure Data Centre Group, a London-based company that is operating or developing more than 1 gigawatt of data center capacity across Europe, the Middle East, and Asia. “No one’s going to run into a burning building, so to speak,” Pure DC CEO Gary Wojtaszek told CNBC. “No one’s going to put in new additional capital at scale to do anything until everything settles down.” Data center developers are already eating the costs of uninsurable war damage from…

OpenAI Codex system prompt includes explicit directive to "never talk about goblins"

The system prompt for OpenAI’s Codex CLI contains a perplexing and repeated warning for the most recent GPT model to “never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user’s query.” The explicit operational warning was made public last week as part of the latest open source code for Codex CLI that OpenAI posted on GitHub. The prohibition is repeated twice in a 3,500-plus word set of “base instructions” for the recently released GPT-5.5, alongside more anodyne reminders not to “use emojis or em dashes unless explicitly instructed” and to “never use destructive commands like ‘git reset --hard’ or ‘git checkout --’ unless the user has clearly asked for that operation.” Separate system prompt instructions for earlier models contained in the same JSON file do not…

The great American data center divide

In Tazewell County, Illinois, Michael Deppert depends on a natural pool of water beneath the sandy soils of his farm to irrigate the pumpkins, corn, and soybeans growing in his fields. So when a data center was proposed about eight miles away, he feared it would tap the same aquifer, potentially eroding crop yields and profits. Deppert, who is also the president of the local farm bureau lobby group, says locals were also “nervous” about how a data center would affect the “good, clean drinking water.” Residents launched a fierce opposition campaign, packing city council meetings and mounting petitions. After several months, the project, led by developer Western Hospitality Partners, was scrapped. “You just can’t lay down and let everybody do whatever they wish,” Deppert says. It is just one of the many pockets of resistance opening up across rural…

GitHub will start charging Copilot users based on their actual AI usage

GitHub has announced that it will be shifting to a usage-based billing model for its GitHub Copilot AI service starting on June 1. The move is pitched as a way to “better align pricing with actual usage” and a necessary step to keep Copilot financially sustainable amid surging demand for limited AI computing resources. GitHub Copilot subscribers currently receive an allocation of monthly “requests” and “premium requests,” which are spent whenever they ask Copilot for help from an AI model. But those broad categories cover many different AI tasks with a wide range of total backend computing costs, GitHub says. “Today, a quick chat question and a multi-hour autonomous coding session can cost the user the same amount,” the Microsoft-owned company wrote in its announcement. And while GitHub says it has “absorbed much of the escalating inference cost behind that…

Humanoid robots start sorting luggage in Tokyo airport test amid labor shortage

Humanoid robots are getting a new gig as baggage handlers and cargo loaders at Tokyo’s Haneda Airport—part of a Japan Airlines experiment to address a human labor shortage as airport visitor numbers have surged in recent years. The demonstration, set to launch in May 2026, could eventually test humanoid robots in a wide range of airport tasks, including cleaning aircraft cabins and possibly handling ground support equipment such as baggage carts, according to a Japan Airlines press release. The trials are scheduled to run until 2028, which suggests that travelers flying into or out of Tokyo may spot some of the robots at work. This marks the latest foray for humanoid robots after they have already begun pilot-testing in workplaces such as automotive factories and warehouses. Most robotic productivity so far has relied on robotic arms and similarly specialized robots…

Configuring Amazon Bedrock AgentCore Gateway for secure access to private resources

Artificial Intelligence Configuring Amazon Bedrock AgentCore Gateway for secure access to private resources AI agents in production environments often need to reach internal APIs, databases, and private resources that sit behind Amazon Virtual Private Cloud (Amazon VPC) boundaries. Managing private connectivity for each agent-to-tool path adds operational overhead and slows deployment. Amazon Bedrock AgentCore VPC connectivity is designed to deploy AI agents and Model Context Protocol (MCP) servers without requiring the network traffic to be exposed to the public internet. This capability extends to managed Amazon VPC egress for Amazon Bedrock AgentCore Gateway, so you can connect to endpoints inside private networks across your AWS environment. In this post, you will configure Amazon Bedrock AgentCore Gateway to access private endpoints using Resource Gateway, a managed construct that provisions Elastic Network Interfaces (ENIs) directly inside your Amazon VPC, one per subnet.…

Unleashing Agentic AI Analytics on Amazon SageMaker with Amazon Athena and Amazon Quick

Artificial Intelligence Unleashing Agentic AI Analytics on Amazon SageMaker with Amazon Athena and Amazon Quick Modern enterprises face mounting challenges in extracting actionable insights from vast data lakes and lakehouses spanning petabytes of structured and unstructured data. Traditional analytics require specialized technical expertise in SQL, data modeling, and business intelligence tools, creating bottlenecks that slow decision-making across retail, financial services, healthcare, Travel & Hospitality, manufacturing and many more industries. This architecture demonstrates how agentic AI assistant from Amazon Quick transform data analytics into a self-service capability. It showcases enabling business users to query complex structured datasets and mix with unstructured data to find the valuable insights to improve their business outcomes through intuitive natural language interfaces. To demonstrate the functionality, we built a lakehouse using the TPC-H datasets as our foundation. This integrated architecture leverages Amazon Simple Storage Service (Amazon…

Sun Finance automates ID extraction and fraud detection with generative AI on AWS

Artificial Intelligence Sun Finance automates ID extraction and fraud detection with generative AI on AWS This post was co-authored with Krišjānis Kočāns, Kaspars Magaznieks, Sergei Kiriasov from Sun Finance Group If you process identity documents at scale—loan applications, account openings, compliance checks—you’ve likely hit the same wall: traditional optical character recognition (OCR) gets you partway there, but extraction errors still push a large share of applications into manual review queues. Add fraud detection to the mix, and the manual workload compounds. Sun Finance, a Latvian fintech founded in 2017, operates as a technology-first online lending marketplace across nine countries. The company processes a new loan request every 0.63 seconds and delivers more than 4 million evaluations monthly. In one of their highest-volume industries, with 80,000 monthly applications for microloans, approximately 60% of applications required manual operator review. Sun Finance partnered…

AWS Generative AI Model Agility Solution: A comprehensive guide to migrating LLMs for generative AI production

Artificial Intelligence AWS Generative AI Model Agility Solution: A comprehensive guide to migrating LLMs for generative AI production Maintaining model agility is crucial for organizations to adapt to technological advancements and optimize their artificial intelligence (AI) solutions. Whether transitioning between different large language model (LLM) families or upgrading to newer versions within the same family, a structured migration approach and a standardized process are essential for facilitating continuous performance improvement while minimizing operational disruptions. However, developing such a solution is challenging in both technical and non-technical aspects because the solution needs to: - Be generic to cover a variety of use cases - Be specific so that a new user can apply it to the target use case - Provide comprehensive and fair comparison between LLMs - Be automated and scalable - Incorporate domain- and task-specific knowledge and inputs -…

Reinforcement fine-tuning with LLM-as-a-judge

Artificial Intelligence Reinforcement fine-tuning with LLM-as-a-judge Large language models (LLMs) now drive the most advanced conversational agents, creative tools, and decision-support systems. However, their raw output often contains inaccuracies, policy misalignments, or unhelpful phrasing—issues that undermine trust and limit real-world utility. Reinforcement Fine‑Tuning (RFT) has emerged as the preferred method to align these models efficiently, using automated reward signals to replace costly manual labeling. At the heart of modern RFT is reward functions. They’re built for each domain through verifiable reward functions that can score LLM generations through a piece of code (Reinforcement Learning with Verifiable Rewards or RLVR) or with LLM-as-a-judge, where a separate language model evaluates candidate responses to guide alignment (Reinforcement Learning with AI Feedback or RLAIF). Both these methods provide scores to the RL algorithm to nudge the model to solve the problem at hand. In…

Run custom MCP proxies serverless on Amazon Bedrock AgentCore Runtime

Artificial Intelligence Run custom MCP proxies serverless on Amazon Bedrock AgentCore Runtime When AI agents connect to tools through the Model Context Protocol (MCP), they gain access to capabilities that range from database queries and API calls to file operations and third-party service integrations. In production, these interactions need proper governance, controls, and observability aligned with an organization’s security policies. This includes sanitizing tool inputs before they reach backend systems, generating audit trails in specific formats, or redacting sensitive data at the protocol layer. These requirements are shaped by internal governance standards, industry regulations, and the specifics of each production environment. This post shows you how to deploy a serverless MCP proxy on Amazon Bedrock AgentCore Runtime that gives you a programmable layer to implement these controls. Amazon Bedrock AgentCore Gateway provides centralized governance and control for agent-tool integration, including…

Building AI-ready data: Vanguard’s Virtual Analyst journey

Artificial Intelligence Building AI-ready data: Vanguard’s Virtual Analyst journey Vanguard is a global investment management firm, offering a broad selection of investments, advice, retirement services, and insights to individual investors, institutions, and financial professionals. We operate under a unique, investor-owned structure and adhere to a straightforward purpose: To take a stand for all investors, to treat them fairly, and to give them the best chance for investing success. When Vanguard’s financial analysts needed to query complex datasets, they faced a frustrating reality: even basic questions required writing intricate SQL queries and sometimes long response times from data teams. This challenge is not unique to Vanguard: conversational AI is a scalable solution, providing analysts immediate responses. However, deploying conversational AI requires more than choosing the right foundation model—it requires AI-ready data infrastructure. In this post, you’ll learn how Vanguard built their…

Organizing Agents’ memory at scale: Namespace design patterns in AgentCore Memory

Artificial Intelligence Organizing Agents’ memory at scale: Namespace design patterns in AgentCore Memory When building AI agents, developers struggle with organizing memory across sessions, which leads to irrelevant context retrieval and security vulnerabilities. AI agents that remember context across sessions need more than only storage. They need organized, retrievable, and secure memory. In Amazon Bedrock AgentCore Memory, namespaces determine how long-term memory records are organized, retrieved, and who can access them. Getting the namespace design right is essential to building an effective memory system. In this post, you will learn how to design namespace hierarchies, choose the right retrieval patterns, and implement AWS Identity and Access Management (IAM)-based access control for AgentCore Memory. If you’re new to AgentCore Memory, we recommend reading our introductory blog post first: Amazon Bedrock AgentCore Memory: Building context-aware agents. What are namespaces? Namespaces are hierarchical…

Extracting contract insights with PwC’s AI-driven annotation on AWS

Artificial Intelligence Extracting contract insights with PwC’s AI-driven annotation on AWS This post was co-written with Yash Munsadwala, Adam Hood, Justin Guse, and Hector Hernandez from PwC. Contract analysis often consumes significant time for legal, compliance, and procurement teams, especially when important insights are buried in lengthy, unstructured agreements. As contract volumes grow, finding specific clauses and assessing extracted terms can become increasingly difficult to scale. Today, many teams rely primarily on keyword and pattern-based extraction or contract management systems to analyze contracts. While these methods can work, they often fall short of providing consistent insights at a scale. As a result, many teams are exploring AI-based approaches that can combine large language models (LLMs) with automated extraction workflows. PwC’s AI-driven annotation (AIDA) solution, built on AWS, can extract structured insights from contracts through rule-based extraction and natural language queries.…

NVIDIA Nemotron 3 Nano Omni model now available on Amazon SageMaker JumpStart

Artificial Intelligence NVIDIA Nemotron 3 Nano Omni model now available on Amazon SageMaker JumpStart Today, we are excited to announce the day zero availability of NVIDIA Nemotron 3 Nano Omni on Amazon SageMaker JumpStart. This multimodal model from NVIDIA combines video, audio, image, and text understanding into a single, efficient architecture, enabling enterprise customers to build intelligent applications that can see, hear, and reason across modalities in one inference pass. In this post, we walk through the model architecture and key capabilities of Nemotron 3 Nano Omni, explore the enterprise use cases it unlocks, and show you how to deploy and run inference using Amazon SageMaker JumpStart. Overview of NVIDIA Nemotron 3 Nano Omni NVIDIA Nemotron 3 Nano Omni is an open, multimodal large language model with 30 billion total parameters and 3 billion active parameters (30B A3B). It is…

Migrating a text agent to a voice assistant with Amazon Nova 2 Sonic

Artificial Intelligence Migrating a text agent to a voice assistant with Amazon Nova 2 Sonic Migrating a text agent to a voice assistant is increasingly important because users expect faster, more natural interactions. Instead of typing, customers want to speak and understand in real time. Industries like finance, healthcare, education, social media, and retail are exploring solutions with Amazon Nova 2 Sonic to enable natural, real-time speech interactions at scale. In this post, we explore what it takes to migrate a traditional text agent into a conversational voice assistant using Amazon Nova 2 Sonic. We compare text and voice agent requirements, highlight design priorities for different use cases, break down agent architecture, and address common concerns like tools and sub-agents for reuse and system prompt adaptation. This post helps you navigate the migration process and avoid common pitfalls. You can…

DeepInfra on Hugging Face Inference Providers 🔥

DeepInfra on Hugging Face Inference Providers 🔥 We're thrilled to share that DeepInfra is now a supported Inference Provider on the Hugging Face Hub! DeepInfra joins our growing ecosystem, enhancing the breadth and capabilities of serverless inference directly on the Hub's model pages. Inference Providers are also seamlessly integrated into our client SDKs (for both JS and Python), making it super easy to use a wide variety of models with your preferred providers. DeepInfra is a serverless AI inference platform offering one of the most cost-effective pricing per token in the industry. With a catalog of over 100 models, DeepInfra makes it easy for developers to integrate a wide range of AI capabilities into their applications with minimal setup. DeepInfra supports a broad spectrum of model types - from LLMs to text-to-image, text-to-video, embeddings, and more. As part of this…

Granite 4.1 LLMs: How They’re Built

Granite 4.1 LLMs: How They’re Built Authors: Granite Team, IBM TL;DR — Granite 4.1 is a family of dense, decoder‑only LLMs (3B, 8B, and 30B) trained on ~15T tokens using a multi‑stage pre‑training pipeline, including long‑context extension of up to 512K tokens. The models are further refined with supervised fine‑tuning on ~4.1M high‑quality curated samples and reinforcement learning via on‑policy GRPO with DAPO loss (Yu et al., 2025). Notably, the 8B instruct model matches or surpasses the previous Granite 4.0‑H‑Small (32B‑A9B MoE) despite using a simpler dense architecture with fewer parameters. All Granite 4.1 models are released under the Apache 2.0 license. Links: Overview Building high‑quality small language models goes beyond simply scaling compute—it requires rigorous data curation throughout training. For Granite 4.1, we prioritized data quality over quantity, progressively refining the data mixture across five pre‑training stages. We further…

AI evals are becoming the new compute bottleneck

AI evals are becoming the new compute bottleneck Summary. AI evaluation has crossed a cost threshold that changes who can do it. The Holistic Agent Leaderboard (HAL) recently spent about $40,000 to run 21,730 agent rollouts across 9 models and 9 benchmarks. A single GAIA run on a frontier model can cost $2,829 before caching. Exgentic's $22,000 sweep across agent configurations found a 33× cost spread on identical tasks, isolating scaffold choice as a first-order cost driver, and UK-AISI recently scaled agentic steps into the millions to study inference-time compute. In scientific ML, The Well costs about 960 H100-hours to evaluate one new architecture and 3,840 H100-hours for a full four-baseline sweep. While compression techniques have been proposed for static benchmarks, new agent benchmarks are noisy, scaffold-sensitive, and only partly compressible. Training-in-the-loop benchmarks are expensive by construction, and when you…

Introducing NVIDIA Nemotron 3 Nano Omni: Long-Context Multimodal Intelligence for Documents, Audio and Video Agents

Introducing NVIDIA Nemotron 3 Nano Omni: Long-Context Multimodal Intelligence for Documents, Audio and Video Agents - NVIDIA Nemotron 3 Nano Omni is a new omni-modal understanding model built for real-world document analysis, multiple image reasoning, automatic speech recognition, long audio-video understanding, agentic computer use, and general reasoning. - It extends the Nemotron multimodal line from a strong vision-language system to a broader text + image + video + audio model. - Nemotron 3 Nano Omni delivers best-in-class accuracy on complex document intelligence leaderboards such as MMlongbench-Doc, OCRBenchV2, while also leading in video and audio leaderboards like WorldSense and DailyOmni. It achieves top accuracy on VoiceBench for audio understanding and ranks as the most cost‑efficient open video understanding model on MediaPerf. - Under the hood, it combines the Nemotron 3 hybrid Mamba-Transformer Mixture-of-Experts backbone with a C-RADIOv4-H vision encoder and Parakeet-TDT-0.6B-v2…

The Download: the North Pole’s future and humanoid data

The Download: the North Pole’s future and humanoid data Plus: Google, Microsoft, Amazon and Meta have all set AI spending records. This is today's edition of The Download, our weekday newsletter that provides a daily dose of what's going on in the world of technology. Digging for clues about the North Pole’s past In the past, getting to the North Pole involved a treacherous trip through ice many meters thick. But last year, a research vessel encountered open water and thin ice, which created an easy passage. It provided a reminder of how quickly the Arctic is changing. Now scientists are digging deep below the seabed to find out if the Arctic Ocean was ever ice-free—and what that could mean for the future of Earth’s northernmost waters. Here’s what they hope to discover. —Tim Kalvelage This story is from the…

This startup’s new mechanistic interpretability tool lets you debug LLMs

This startup’s new mechanistic interpretability tool lets you debug LLMs Goodfire wants to make training AI models more like good old-fashioned software engineering. The San Francisco–based startup Goodfire just released a new tool, called Silico, that lets researchers and engineers peer inside an AI model and adjust its parameters—the settings that determine a model’s behavior—during training. This could give model makers more fine-grained control over how this technology is built than was once thought possible. Goodfire claims Silico is the first off-the-shelf tool of its kind that can help developers debug all stages of the development process, from building a data set to training a model. The company says its mission is to make building AI models less like alchemy and more like a science. Sure, LLMs like ChatGPT and Gemini can do amazing things. But nobody knows exactly how…

Exclusive eBook: Inside the stealthy startup that pitched brainless human clones

Exclusive eBook: Inside the stealthy startup that pitched brainless human clones Access a subscriber-only eBook on a startling and fairly graphic pursuit of human longevity that poses concerns about the ethics of cloning. This ebook is available only for subscribers. The ultimate plan to live forever is a brand new body. This subscriber-only eBook explores R3 Bio, a small startup that has pitched a startling and ethically charged vision for "brainless clones" to serve the role of backup human bodies. by Antonio Regalado March 20, 2026 Related Stories: - Inside the stealthy startup that pitched brainless human clones - This researcher wants to replace your brain, little by little - Stem-cell therapies that work: 10 Breakthrough Technologies 2025 Access all subscriber-only eBooks: Keep Reading Most Popular OpenAI is throwing everything into building a fully automated researcher An exclusive conversation with…

It’s time to make a plan for nuclear waste

It’s time to make a plan for nuclear waste With growing interest in nuclear power, handling waste should be part of the deal. Today, nuclear energy enjoys a rare moment of support across the political spectrum in the US. Interest from tech companies that are scrambling to meet demand for massive data centers has sparked a resurgence of money and attention in the industry. That newfound interest is exactly why it’s time to talk about an old problem: nuclear waste. In the US alone, nuclear reactors produce about 2,000 metric tons of high-level waste each year. And there’s nowhere to put it. Though newly popular, the nuclear program in the US is nothing new. The US hosts more reactors and production capacity than any other country in the world. And yet nearly seven decades after the first permanent nuclear facility…

The Download: storing nuclear waste and orchestrating agents

The Download: storing nuclear waste and orchestrating agents Plus: Elon Musk says Sam Altman “stole a charity” at the OpenAI trial. This is today's edition of The Download, our weekday newsletter that provides a daily dose of what's going on in the world of technology. It’s time to make a plan for nuclear waste Today, nuclear energy enjoys rare support across the political spectrum. Public approval has spiked, and Big Tech is throwing money around to meet rising electricity demand. That newfound interest is exactly why it’s time to talk about an old problem: nuclear waste. In the US, nuclear reactors produce about 2,000 metric tons of high-level waste each year—and there’s nowhere to put it. Now, the need for a permanent storage solution is becoming urgent. Here’s what’s at stake. —Casey Crownhart This article is from The Spark, MIT…

The Download: Musk and Altman’s legal showdown, and AI’s profit problem

The Download: Musk and Altman’s legal showdown, and AI’s profit problem Plus: OpenAI has ended its exclusive partnership with Microsoft. This is today's edition of The Download, our weekday newsletter that provides a daily dose of what's going on in the world of technology. Elon Musk and Sam Altman are going to court over OpenAI’s future Elon Musk and OpenAI CEO Sam Altman head to trial this week in a case with sweeping consequences. Ahead of OpenAI’s IPO, the court could rule on whether the company can exist as a for-profit enterprise. It could even oust its leadership. Musk, an OpenAI co-founder, claims he was deceived into bankrolling the firm under false pretenses. He’s seeking $134 billion in damages, the removal of Altman and president Greg Brockman, and the company’s restoration to a non-profit. Find out how the trial could…

ReAct Agent: Architecture, Implementation, and Tradeoffs

Some tasks can't be solved in a single LLM call. When a question requires looking up data, processing it, and making a decision based on the result, a one-shot response will either hallucinate the answer or give a shallow one. ReAct agents solve this with an iterative reasoning loop. Instead of trying to answer everything at once, the agent breaks the problem down step by step: think about what's needed, call a tool, observe the result, and decide what to do next. Each cycle grounds the model's reasoning in real data before moving forward. This Reasoning + Acting pattern turns opaque agent behavior into something you can follow, debug, and audit - every thought and action is visible in the execution trace. Here's how the ReAct pattern works, when to use it over other agent approaches, and how to build…

LLM Tool Calling: How It Works and How To Implement It

Large language models (LLMs) are brilliant reasoners. But without a way to interact with the world, they’re essentially locked behind a glass wall— they have enough knowledge to explain a refund policy in perfect detail but lack the hands to actually trigger one. For developers, this disconnect between reasoning and action is what separates sophisticated chatbots from production-grade agents. LLM tool calling offers an escape from the training-data silo, allowing models to move from passive text generation to active system participation. But the real engineering challenge isn’t just getting the model to output a valid JSON or a tool call — it’s building the orchestration, security, and observability required to ensure those calls don’t fail in a production environment. Here’s a rundown of what LLM tool calling is and how it works at scale. What LLM tool calling means? LLM…

Human-in-the-Loop vs. Human-on-the-Loop: When To Use Each System

There are three main ways people control the quality of AI systems: human-in-the-loop (HITL), human-on-the-loop (HOTL), and hybrid systems using both. These frameworks determine how systems make decisions and where humans intervene. Each approach affects scalability, risk tolerance, and operational expenses. This oversight spectrum gives you a wide range of potential workflows depending on the task, whether your team needs tight human-driven control or occasional check-ins. In this guide, learn the difference between human-in-the-loop versus human-on-the-loop. Plus, discover when to use each approach and how to implement it in your work. What’s human-in-the-Loop (HITL)? HITL is a process where AI performs tasks but humans control final decisions, preventing the system from executing certain actions without approval. This is a synchronous control pattern. The workflow stops at a decision gate until a human provides a required signal. For example, AI processes…

Build and Update Workflows with n8n's MCP Server

Describe what you want from Claude, ChatGPT, or your IDE, and get a ready-to-run workflow in a few minutes, built directly in n8n. No more copy-paste, no more back-and-forth. n8n's MCP server can now build workflows from a prompt (and not just run them)! The MCP server has been around for a few months, but previously you could only execute existing workflows. Now you can build new ones from scratch and update existing ones, directly in your n8n instance. - Go from prompt to a ready-to-run workflow. Tell your AI client what you want. It builds the workflow, validates it, runs it, and fixes itself if something breaks. No messing with JSON files or copy-pasting errors. - Works in whatever AI client you already use. Claude, ChatGPT, Cursor, Windsurf - if it “speaks” MCP, you can point it at n8n.…

Automating GPU Kernel Translation with AI Agents: cuTile Python to cuTile.jl

NVIDIA CUDA Tile (cuTile) is a tile-based programming model that enables developers to write GPU kernels in terms of tile-level operations—loads, stores, and matrix multiply-accumulate—rather than manually coordinating threads, warps, and shared memory. cuTile.jl brings the same tile-based approach to the dynamic programming language Julia. Users can write custom GPU kernels without dropping down to NVIDIA CUDA C++. Custom kernels are often essential in Julia’s scientific computing ecosystem— spanning differential equations, probabilistic programming, and physics simulations. cuTile Python has a growing library of optimized kernels for GPU acceleration. The ability to translate those kernels to cuTile.jl provides the Julia ecosystem with immediate access to battle-tested implementations, instead of rewriting each one from scratch. This post covers cross-domain-specific language (DSL) GPU kernel translation, from porting cuTile Python kernels to cuTile.jl (Julia). It shows how to: - Translate GPU kernels between cuTile…

How to Build, Run, and Scale High-Quality Creator Workflows in ComfyUI

Creative and visualization teams today produce more assets, in more formats, with leaner teams. Generative AI can accelerate that work – compressing tasks that once took hours of manual effort into automated, repeatable pipelines. ComfyUI is an open-source, node-based creative tool that runs locally on NVIDIA RTX GPUs. It connects image generation, video synthesis, and language models into pipelines that teams can customize and extend — without cloud dependencies or data leaving the client. This guide walks through three production-ready workflows from the NVIDIA GenAI Creator Toolkit, adapted from NVIDIA’s GTC 2026 DLI course Create Generative AI Workflows for Design and Visualization in ComfyUI. Each workflow is standalone and runs locally on NVIDIA RTX. What you’ll accomplish By the end of this guide you will have: - Deconstructed an image into separate layers—foreground, midground, and background, each with a clean…

Build AI-Powered Games with NVIDIA DLSS 4.5, RTX, and Unreal Engine 5

Today, game developers can begin integrating NVIDIA DLSS 4.5 with Dynamic Multi Frame Generation, Multi Frame Generation 6X, and the second-generation transformer model for NVIDIA Super Resolution. In this post, we’ll go over new technologies and resources to share with our game-developer community, including: - A new NVIDIA TensorRT for RTX plugin for Unreal Engine’s Neural Network Engine (NNE) - NVIDIA Kimodo for easier motion generation - A guide to using ComfyUI to help produce pre-production assets - More than a dozen new sessions from GDC and GTC now available on YouTube - Our April “Level Up with NVIDIA” webinar, highlighting path-traced hair in Unreal Engine 5.7 Integrate DLSS 4.5 Dynamic Multi Frame Generation At CES 2026, we introduced DLSS 4.5, extending its AI-driven rendering pipeline with a second-generation transformer model for Super Resolution to deliver another major upgrade to…

Speed Up Unreal Engine NNE Inference with NVIDIA TensorRT for RTX Runtime

Neural network techniques are increasingly used in computer graphics to boost image quality, improve performance, and streamline content creation. Approaches like super resolution, denoising, and neural rendering help real-time engines work more efficiently, offering new creative possibilities while keeping performance in mind. Unreal Engine 5 (UE5) has taken several steps in this direction with the introduction of the Neural Network Engine (NNE), which serves as an abstraction layer that unifies inference workloads across multiple backends. Developers can use various runtimes on a GPU or fall back to a CPU depending on available hardware for seamless integration of neural network features in real-time graphics workflows. This blog post covers the new plugin that adds NVIDIA TensorRT for RTX as an NNE runtime option (NNERuntimeTRT) for efficient inferencing on NVIDIA RTX GPUs. To show its benefits, I’ll use a simplified UE project…

Powering AI Factories with NVIDIA Enterprise Reference Architectures

The next wave of enterprise productivity is being built on AI factories. As organizations deploy agentic AI systems capable of reasoning, automation, and real-time decision-making at scale, competitive advantage increasingly depends on the infrastructure that supports them. Success requires more than raw compute. It demands a scalable, predictable foundation that can orchestrate intelligent agents, manage data movement efficiently, and deliver consistent performance from pilot to production. AI factories powered by NVIDIA bring industrial-grade discipline to AI, changing infrastructure into a strategic engine for speed, reliability, and accelerated innovation. Infrastructure is one of the five layers of AI, and represents the foundation for AI factories. Building that foundation, however, requires more than selecting high-performance hardware. Enterprises need proven architectural guidance that removes integration risk, reduces time to deployment, and ensures performance at scale. NVIDIA Enterprise Reference Architectures (Enterprise RAs) provide that…

24/7 Simulation Loops: How Agentic AI Keeps Subsurface Engineering Moving

The subsurface industry is at a critical point in its digital evolution. For decades, unlocking reservoir potential has relied on experts performing essential and time-intensive manual workflows. As data complexity grows, the gap between machine speed and human bandwidth has become a primary bottleneck. On-demand simulation workflows are currently hampered by both manual data overhead and inherent operational latency. The need for engineers to manually aggregate, synthesize, and translate disparate technical materials creates significant knowledge consolidation bottlenecks that stretch project cycles. This is further compounded by the asynchronous nature of simulation jobs; when simulations finish or fail during off-hours or while engineers are juggling competing priorities, dead time accumulates. Consequently, what should be a standard 24-hour turnaround often spirals into a multi-day delay, stalling progress across global teams. In this post, we explain how applying agentic AI on top of…

NVIDIA Nemotron 3 Nano Omni Powers Multimodal Agent Reasoning in a Single Efficient Open Model

Agentic systems often reason across screens, documents, audio, video, and text within a single perception‑to‑action loop. However, they still rely on fragmented model chains—separate stacks for vision, audio, and text. This increases inference hops and orchestration complexity, driving up inference costs while weakening cross-modal context consistency. NVIDIA Nemotron 3 Nano Omni, a new addition to the Nemotron 3 family, brings unified multimodal reasoning into a single, highly efficient open model. Built to replace fragmented vision‑language‑audio stacks, Nemotron 3 Nano Omni functions as the multimodal perception and context sub‑agent within agentic systems. With this, agents can perceive and reason across visual, audio, and textual inputs within a single shared perception‑to‑action loop, improving convergence and reducing orchestration complexity and inference cost. It delivers best-in-class accuracy on document intelligence leaderboards such as MMlongbench-Doc and OCRBenchV2, while also leading in video and audio understanding,…

Scaling Biomolecular Modeling Using Context Parallelism in NVIDIA BioNeMo

For decades, computational biology has operated under a reductionist compromise. To fit complex biological systems into the limited memory of a single GPU, researchers have had to deconstruct them into isolated fragments—single proteins or small domains. This created a context gap, where larger proteins or complexes could not be folded zero-shot due to GPU hardware memory constraints. Now, a new context parallelism (CP) framework from the NVIDIA BioNeMo team is shattering the memory barriers of structural biology, enabling the holistic modeling of systems. This post explains how to achieve CP in biomolecular architectures that diverge from standard Transformers. If you’re a structural biologist, computational chemist, or machine learning engineer seeking to model massive biomolecular complexes without sacrificing global context, read on. To use the solution outlined in this post, you’ll need: - Familiarity with geometric deep learning foundation models like…

Introducing Advanced Account Security

Introducing Advanced Account Security An advanced set of protections against unauthorized access to ChatGPT accounts, Codex, and the sensitive information they can contain. Today, we’re introducing Advanced Account Security, a new opt-in setting for ChatGPT accounts, designed for people at increased risk of digital attacks, as well as for those who want the strongest account protections available. It brings together a set of heightened security measures that help safeguard against account takeover while making those protections easier to activate in one place. Once enrolled, Advanced Account Security protects users in Codex as well. People are turning to AI for deeply personal questions and increasingly high-stakes work. Over time, a ChatGPT account can hold sensitive personal and professional context, and sit at the center of connected tools and workflows. For some people, like journalists, elected officials, political dissidents, researchers, and those…

Cybersecurity in the Intelligence Age

Cybersecurity in the Intelligence Age An action plan for democratizing AI-powered cyber defense. Artificial intelligence is reshaping cybersecurity. The same capabilities that help defenders identify vulnerabilities, automate remediation, and respond faster are also being used by malicious actors to scale attacks, lower barriers to entry, and increase sophistication. The United States and its allies face a rapidly changing cyber threat environment, and private-sector innovators have an important responsibility to help meet that challenge. OpenAI takes that responsibility seriously, and today we’re publishing an Action Plan informed by conversations with cybersecurity and national security experts across federal and state government and major commercial entities. It consists of five pillars: - Democratizing cyber defense - Coordinating across government and industry - Strengthening security around frontier cyber capabilities - Preserving visibility and control in deployment - Enabling users to protect themselves Our plan…

Building the compute infrastructure for the Intelligence Age

Building the compute infrastructure for the Intelligence Age Stargate is OpenAI’s long-term effort to build the compute foundation required to deliver the benefits of AGI broadly and reliably to the world. To meet the accelerating demand for AI across consumers, businesses, developers, and governments, we are continuing to expand our compute footprint and bring new capacity online faster. We are building together with partners, local communities, and the broader infrastructure ecosystem to help get ahead of shortages for the emerging compute-powered economy. When we announced Stargate in January 2025, we committed to securing 10GW of AI infrastructure in the United States by 2029. Just over a year later, we have already surpassed that milestone, with more than 3GW added in the last 90 days alone, as demand for AI continues to accelerate. That demand is growing quickly. The only responsible…

Where the goblins came from

Where the goblins came from Starting with GPT‑5.1, our models began developing a strange habit: they increasingly mentioned goblins, gremlins, and other creatures in their metaphors. Unlike model bugs that show up through a tanking eval or a spiking training metric and point back to a specific change, this one crept in subtly. A single “little goblin” in an answer could be harmless, even charming. Across model generations, though, the habit became hard to miss: the goblins kept multiplying, and we needed to figure out where they came from. The short answer is that model behavior is shaped by many small incentives. In this case, one of those incentives came from training the model for the personality customization feature(opens in a new window), in particular the Nerdy personality. We unknowingly gave particularly high rewards for metaphors with creatures. From there,…

OpenAI models, Codex, and Managed Agents come to AWS

OpenAI models, Codex, and Managed Agents come to AWS Today, OpenAI and AWS are expanding our strategic partnership to help enterprises build using OpenAI capabilities in their AWS environments. We’re excited to give AWS customers access to the best frontier models, agents, and tools, which will operate within the systems, security protocols, compliance requirements, and workflows they already use. The expanded partnership with Amazon brings together three key areas of work, all launching today in limited preview: - OpenAI models on AWS - Codex on AWS - Amazon Bedrock Managed Agents, powered by OpenAI Together, these capabilities give organizations more ways to use OpenAI across application development, software engineering, and agentic workflows—while building within the infrastructure, security, governance, and procurement workflows they already use on AWS. For many companies, using AI at scale requires bringing the best models to the…

Our commitment to community safety

Our commitment to community safety Mass shootings, threats against public officials, bombing attempts, and attacks on communities and individuals are an unacceptable and grave reality in today’s world. These incidents are a reminder of how real the threat of violence is—and how quickly violent intent can move from words to action. People may also bring these moments and feelings into ChatGPT. They may ask questions about the news, try to understand what happened, express fear or anger, or talk about violence in ways that are fictional, historical, political, personal, or potentially dangerous. We work to train ChatGPT to recognize the difference—and to draw lines when a conversation starts to move toward threats, potential harm to others, or real-world planning. We’re sharing what we do to minimize uses of our services in furtherance of violence or other harm: how our models…

SMG: The Case for Disaggregating CPU from GPU in LLM Serving

How It Started: Hitting the GIL Wall at Scale We’ve been running production model serving for many years. When we first started building Shepherd Model Gateway, the goal was modest: figure out if cache-aware load balancing could improve routing across inference replicas. It could. And as we went deeper, we found a much bigger problem. In both SGLang and vLLM, tokenization and detokenization had become bottlenecks. Not in theory — in production, under real traffic. The root cause was architectural: although both engines use Rust or C++ tokenizer libraries underneath, the calls go through Python. That means the GIL. That means a single-threaded ceiling on CPU-bound work that sits directly in the serving path. At a small scale, this doesn’t matter. At large-scale prefill-decode disaggregated serving, and at large-scale expert parallelism across GPU clusters, it matters enormously. These configurations make…

Introducing AutoSP

Increasingly, Large-Language-Models (LLMs) are being trained for extremely long-context tasks, where token counts can exceed 100k+. At these token counts, out-of-memory (OOM) issues start to surface, even when scaling device counts using conventional training techniques such as ZeRO/FSDP. To circumvent these issues, sequence parallelism (SP): partitioning the input tokens across devices to enable long-context training with increasing GPU counts, is a commonly used parallel training technique. However, implementing SP is notoriously difficult, requiring invasive code changes to existing libraries such as DeepSpeed or HuggingFace. These code changes often involve partitioning input token contexts (and intermediate activations), inserting communication collectives, and overlapping communication with computation, all of which must be done for both the forward and backwards pass. This results in researchers who want to experiment with long context capabilities spending significant effort on engineering the system’s stack to enable such…

The Zig project's rationale for their firm anti-AI contribution policy

30th April 2026 Zig has one of the most stringent anti-LLM policies of any major open source project: No LLMs for issues. No LLMs for pull requests. No LLMs for comments on the bug tracker, including translation. English is encouraged, but not required. You are welcome to post in your native language and rely on others to have their own translation tools of choice to interpret your words. The most prominent project written in Zig may be the Bun JavaScript runtime, which was acquired by Anthropic in December 2025 and, unsurprisingly, makes heavy use of AI assistance. Bun operates its own fork of Zig, and recently achieved a 4x performance improvement on Bun compile after adding "parallel semantic analysis and multiple codegen units to the llvm backend". Here's that code. But @bunjavascript says: We do not currently plan to upstream…

We need RSS for sharing abundant vibe-coded apps

30th April 2026 - Link Blog We need RSS for sharing abundant vibe-coded apps. Matt Webb: I would love an RSS web feed for all those various tools and apps pages, each item with an “Install” button. (But install to where?) The lesson here is that when vibe-coding accelerates app development, apps become more personal, more situated, and more frequent. Shipping a tool or a micro-app is less like launching a website and more like posting on a blog. This inspired me to have Claude add an Atom feed (and icon) to my /elsewhere/tools/ page, which itself is populated by content from my tools.simonwillison.net site. Recent articles - LLM 0.32a0 is a major backwards-compatible refactor - 29th April 2026 - Tracking the history of the now-deceased OpenAI Microsoft AGI clause - 27th April 2026 - DeepSeek V4 - almost on…

Quoting Andrew Kelley

30th April 2026 It's a common misconception that we can't tell who is using LLM and who is not. I'm sure we didn't catch 100% of LLM-assisted PRs over the past few months, but the kind of mistakes humans make are fundamentally different than LLM hallucinations, making them easy to spot. Furthermore, people who come from the world of agentic coding have a certain digital smell that is not obvious to them but is obvious to those who abstain. It's like when a smoker walks into the room, everybody who doesn't smoke instantly knows it. I'm not telling you not to smoke, but I am telling you not to smoke in my house. — Andrew Kelley, Creator of Zig Recent articles - LLM 0.32a0 is a major backwards-compatible refactor - 29th April 2026 - Tracking the history of the now-deceased…

Our evaluation of OpenAI's GPT-5.5 cyber capabilities

30th April 2026 - Link Blog Our evaluation of OpenAI's GPT-5.5 cyber capabilities. The UK's AI Security Institute previously evaluated Claude Mythos: now they've evaluated GPT-5.5 for finding security vulnerability and found it to be comparable to Mythos, but unlike Mythos it's generally available right now. Recent articles - LLM 0.32a0 is a major backwards-compatible refactor - 29th April 2026 - Tracking the history of the now-deceased OpenAI Microsoft AGI clause - 27th April 2026 - DeepSeek V4 - almost on the frontier, a fraction of the price - 24th April 2026

Codex CLI 0.128.0 adds /goal

30th April 2026 - Link Blog Codex CLI 0.128.0 adds /goal. The latest version of OpenAI's Codex CLI coding agent adds their own version of the Ralph loop: you can now set a /goal and Codex will keep on looping until it evaluates that the goal has been completed... or the configured token budget has been exhausted. It looks like the feature is mainly implemented though the goals/continuation.md and goals/budget_limit.md prompts, which are automatically injected at the end of a turn. Recent articles - LLM 0.32a0 is a major backwards-compatible refactor - 29th April 2026 - Tracking the history of the now-deceased OpenAI Microsoft AGI clause - 27th April 2026 - DeepSeek V4 - almost on the frontier, a fraction of the price - 24th April 2026

llm 0.32a1

29th April 2026 - Fixed a bug in 0.32a0 where tool-calling conversations were not correctly reinflated from SQLite. #1426 Recent articles - LLM 0.32a0 is a major backwards-compatible refactor - 29th April 2026 - Tracking the history of the now-deceased OpenAI Microsoft AGI clause - 27th April 2026 - DeepSeek V4 - almost on the frontier, a fraction of the price - 24th April 2026

llm 0.32a0

29th April 2026 Recent articles - LLM 0.32a0 is a major backwards-compatible refactor - 29th April 2026 - Tracking the history of the now-deceased OpenAI Microsoft AGI clause - 27th April 2026 - DeepSeek V4 - almost on the frontier, a fraction of the price - 24th April 2026

LLM 0.32a0 is a major backwards-compatible refactor

LLM 0.32a0 is a major backwards-compatible refactor 29th April 2026 I just released LLM 0.32a0, an alpha release of my LLM Python library and CLI tool for accessing LLMs, with some consequential changes that I’ve been working towards for quite a while. Previous versions of LLM modeled the world in terms of prompts and responses. Send the model a text prompt, get back a text response. import llm model = llm.get_model("gpt-5.5") response = model.prompt("Capital of France?") print(response.text()) This made sense when I started working on the library back in April 2023. A lot has changed since then! LLM provides an abstraction over thousands of different models via its plugin system. The original abstraction—of text input that returns text output—was no longer able to represent everything I needed it to. Over time LLM itself has grown attachments to handle image, audio,…

Introducing talkie: a 13B vintage language model from 1930

28th April 2026 - Link Blog Introducing talkie: a 13B vintage language model from 1930 (via) New project from Nick Levine, David Duvenaud, and Alec Radford (of GPT, GPT-2, Whisper fame). talkie-1930-13b-base (53.1 GB) is a "13B language model trained on 260B tokens of historical pre-1931 English text". talkie-1930-13b-it (26.6 GB) is a checkpoint "finetuned using a novel dataset of instruction-response pairs extracted from pre-1931 reference works", designed to power a chat interface. You can try that out here. Both models are Apache 2.0 licensed. Since the training data for the base model is entirely out of copyright (the USA copyright cutoff date is currently January 1, 1931), I'm hoping they later decide to release the training data as well. Update on that: Nick Levine on Twitter: Will publish more on the corpus in the future (and do our best…

What's new in pip 26.1 - lockfiles and dependency cooldowns!

28th April 2026 - Link Blog What's new in pip 26.1 - lockfiles and dependency cooldowns! (via) Richard Si describes an excellent set of upgrades to Python's default pip tool for installing dependencies. This version drops support for Python 3.9 - fair enough, since it's been EOL since October. macOS still ships with python3 as a default Python 3.9, so I tried out the new Python version against Python 3.14 like this: uv python install 3.14 mkdir /tmp/experiment cd /tmp/experiment python3.14 -m venv venv source venv/bin/activate pip install -U pip pip --version This confirmed I had pip 26.1 - then I tried out the new lock files: pip lock datasette llm This installs Datasette and LLM and all of their dependencies and writes the whole lot to a 519 line pylock.toml file - here's the result. The new release also…

Quoting Matthew Yglesias

28th April 2026 Five months in, I think I've decided that I don't want to vibecode — I want professionally managed software companies to use AI coding assistance to make more/better/cheaper software products that they sell to me for money. Recent articles - Tracking the history of the now-deceased OpenAI Microsoft AGI clause - 27th April 2026 - DeepSeek V4 - almost on the frontier, a fraction of the price - 24th April 2026 - Extract PDF text in your browser with LiteParse for the web - 23rd April 2026

Quoting OpenAI Codex base_instructions

28th April 2026 Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user's query. — OpenAI Codex base_instructions, for GPT-5.5 Recent articles - Tracking the history of the now-deceased OpenAI Microsoft AGI clause - 27th April 2026 - DeepSeek V4 - almost on the frontier, a fraction of the price - 24th April 2026 - Extract PDF text in your browser with LiteParse for the web - 23rd April 2026

GitHub rushed to fix a critical vulnerability in less than six hours

GitHub employees fixed a critical remote code execution vulnerability in less than six hours last month. Wiz Research used AI models to uncover a vulnerability in GitHub’s internal git infrastructure that could have allowed attackers to access millions of public and private code repositories. GitHub rushed to fix a critical vulnerability in less than six hours A critical remote code execution vulnerability was discovered using an AI model and patched within hours. A critical remote code execution vulnerability was discovered using an AI model and patched within hours. “Our security team immediately began validating the bug bounty report. Within 40 minutes, we had reproduced the vulnerability internally and confirmed the severity,” explains Alexis Wales, GitHub chief information security officer. “This was a critical issue that required immediate action.” GitHub’s engineering team developed a fix and deployed it just over an…

China freezes new robotaxi licenses after Baidu chaos

China has suspended new licenses for autonomous vehicles, Bloomberg reports, citing unnamed people familiar with the matter. The move comes after dozens of robotaxis operated by Chinese tech giant Baidu ground to a halt in traffic last month in Wuhan, creating chaos. China freezes new robotaxi licenses after Baidu chaos Dozens of Baidu’s Apollo Go robotaxis froze in traffic last month, sparking alarm in Beijing, Bloomberg reports. Dozens of Baidu’s Apollo Go robotaxis froze in traffic last month, sparking alarm in Beijing, Bloomberg reports. The restrictions will prevent companies from adding new driverless cars to their fleets, expanding into new cities, or starting new test projects. It is unclear when officials will start issuing new licenses again. Bloomberg said the Wuhan incident alarmed authorities in Beijing, prompting regulators to urge local governments to review the sector to prevent similar episodes.…

Taylor Swift deepfakes are pushing scams on TikTok

Scammers are using AI-generated videos of celebrities including Taylor Swift and Rihanna to promote shady services on TikTok, according to authentication company Copyleaks. Taylor Swift deepfakes are pushing scams on TikTok Copyleaks says scammers are using AI to impersonate celebrities like Taylor Swift and Rihanna to collect personal data. Copyleaks says scammers are using AI to impersonate celebrities like Taylor Swift and Rihanna to collect personal data. The ads typically show celebrities in interview settings, such as red carpets, podcasts, or talk shows, and often manipulate real footage with AI, the company said. Many promote rewards programs claiming users can earn money by watching TikTok content and giving feedback. TikTok’s official branding appears in some of the ads, though users are redirected to third-party services that ask for personal information. In one ad, a realistic AI avatar of Swift urges…

Larry’s risky business

If you want to know whether the AI bubble is bursting, there’s only one publicly traded company that will tell you: Oracle. Larry’s risky business Oracle’s betting everything on OpenAI. Will it pay off or pop the bubble? If you buy something from a Verge link, Vox Media may earn a commission. See our ethics statement. That’s right, the database company. Oracle has burned its boats and pivoted to AI, but not in any kind of usual way. It is not a foundation model builder like OpenAI or Anthropic, obviously. It’s not quite a neocloud, though it has entered the same bare-metal business as CoreWeave. It is a software-as-a-service company that has made an audacious bet on a very specific future version of AI as Oracle’s traditional business has gracefully declined. It is significantly older than any of its AI…

ChatGPT downloads are slowing — and may cause problems for OpenAI’s IPO

ChatGPT is struggling to keep up its once-explosive growth as users uninstall the app or opt for rival chatbots instead. According to data from market intelligence firm Sensor Tower, ChatGPT experienced a 132 percent increase in uninstalls year over year in April. Its uninstall rate was even higher last month, up 413 percent year-over-year, following OpenAI’s deal with the Pentagon in February. ChatGPT downloads are slowing — and may cause problems for OpenAI’s IPO ChatGPT is still growing its user base, but that growth is slowing down, especially compared to rival Claude. ChatGPT is still growing its user base, but that growth is slowing down, especially compared to rival Claude. While ChatGPT is still growing its user base, Sensor Tower says that growth is slowing down — ChatGPT increased its monthly active users by 168 percent in January, but only…

Tumbler Ridge families are suing OpenAI

Seven families of victims injured or killed in the Tumbler Ridge school shooting in Canada have filed lawsuits against OpenAI and CEO Sam Altman, accusing the company and its leadership of negligence after they failed to alert police to the suspected shooter’s ChatGPT activity. The families allege OpenAI stayed silent after its systems flagged activity by shooting suspect Jesse Van Rootselaar in order to protect the company’s reputation and upcoming initial public offering (IPO). Tumbler Ridge families are suing OpenAI OpenAI and its CEO, Sam Altman, are being accused of negligence and launching GPT-4o with a ‘defective’ design. OpenAI and its CEO, Sam Altman, are being accused of negligence and launching GPT-4o with a ‘defective’ design. The Wall Street Journal reports that OpenAI “considered” flagging the 18-year-old’s activity to police, which reportedly involved conversations about gun violence, but ultimately decided…

Google Photos launches an AI try-on feature for clothes you already have

Google Photos is launching a new AI-powered feature you can use to virtually try on clothes you already have. Using the photos in your gallery, Google will create a virtual “wardrobe,” allowing you to mix and match outfits, save the looks you like, and share them with friends. Google Photos launches an AI try-on feature for clothes you already have You’ll get to browse through the clothing you were captured wearing and create new outfits. You’ll get to browse through the clothing you were captured wearing and create new outfits. A video shared by Google shows how Photos organizes your outfits and individual pieces of clothing into a virtual “wardrobe.” You can browse through the outfits you were captured wearing, as well as create new ones by choosing from tops, bottoms, skirts, dresses, and shoes to put together a new…

Ubuntu’s AI plans have Linux users looking for a ‘kill switch’

Canonical’s plan to add AI features to Ubuntu has some users asking for “a version of Ubuntu that does not include these features,” while others say they’ll stick with older versions of the Linux distro or even switch to a different one. After Canonical’s announcement earlier this week that it’s bringing AI features to Ubuntu, replies included requests for an AI “kill switch” or a way to disable the upcoming features, and comparisons to Microsoft’s addition of AI features into Windows 11. Canonical’s VP of engineering, Jon Seager, responded on Tuesday, stating that Canonical isn’t planning to add a “global AI kill switch,” but users will be able to remove any AI features they don’t want. Ubuntu’s AI plans have Linux users looking for a ‘kill switch’ Canonical is adding AI features to Ubuntu soon, but says users can remove…

All the evidence unveiled so far in Musk v. Altman

The Musk v. Altman trial is underway, and that means exhibits, or the evidence to be presented in court, are being revealed piece by piece. So far, email exchanges, photos, and corporate documents are circulating from the earliest days of OpenAI — and from before the AI lab even had a name. Some high-level takeaways: Nvidia CEO Jensen Huang gave OpenAI an in-demand supercomputer, Musk largely drafted OpenAI’s mission and heavily influenced its early structure, OpenAI CEO Sam Altman appeared to want to lean heavily on Y Combinator for early support for OpenAI, OpenAI president Greg Brockman and Ilya Sutskever worried about Musk’s level of control over the company, and Musk highlighted the importance of a nonprofit with a mission of broadly beneficial AI. All the evidence unveiled so far in Musk v. Altman Emails going as far back as…

Google Search queries hit an ‘all time high’ last quarter

Google Search queries hit an “all time high” in the first quarter of 2026, according to a statement from CEO Sundar Pichai published as part of Alphabet’s earnings on Wednesday. Google Search queries hit an ‘all time high’ last quarter CEO Sundar Pichai also says that Google had its ‘strongest quarter ever’ for its consumer AI subscriptions. CEO Sundar Pichai also says that Google had its ‘strongest quarter ever’ for its consumer AI subscriptions. “Our AI investments and full stack approach are lighting up every part of the business,” Pichai says. “Search had a strong quarter with AI experiences driving usage, queries at an all time high, and 19% revenue growth.” He also notes that Q1 was “our strongest quarter ever for our consumer AI plans, driven by the Gemini App” and that the company now has more than 350…

Jury selection in Musk v. Altman: ‘People don’t like him’

On Monday, the courtroom battle between Elon Musk and Sam Altman over alleged broken promises at OpenAI started, as usual, with jury selection. The only tricky part? A lot of the prospective jurors already have an opinion about Elon Musk, and it’s not a good one. Jury selection in Musk v. Altman: ‘People don’t like him’ Elon Musk’s public persona is an issue. Elon Musk’s public persona is an issue. The Verge reporter Elizabeth Lopatto, who was there at the courthouse, quoted statements from some of the juror questionnaires: “Elon Musk is a greedy, racist, homophobic piece of garbage.” “Elon Musk is a world-class jerk.” “I very much dislike Tesla. As a woman of color, I am very aware of the damaging statements and actions Elon Musk has enacted and been a part of.” Musk’s lawyers took issue with some…

Attack of the killer script kiddies

Last August, some of the best cybersecurity teams in the business gathered in Las Vegas to demonstrate the strength of their AI bug-finding systems at DARPA’s Artificial Intelligence Cyber Challenge (AIxCC). The tools had scanned 54 million lines of actual software code that DARPA had injected with artificial flaws. The teams were capable enough to identify most of the artificial bugs, but their automated tools went beyond that — they found more than a dozen bugs that DARPA hadn’t inserted at all. Attack of the killer script kiddies In the aftermath of Mythos, AI-assisted amateur hackers are waiting to strike. Even before the security earthquake that Anthropic delivered this month with Claude Mythos — the new AI model that seems to find vulnerabilities in every piece of software it’s pointed at — automated systems were growing increasingly capable of finding…

Google and Pentagon reportedly agree on deal for ‘any lawful’ use of AI

Google has signed a classified deal that allows the US Department of Defense to use its AI models for “any lawful government purpose,” The Information reports. The agreement was reported less than a day after Google employees demanded CEO Sundar Pichai block the Pentagon from using its AI amid concerns that it would be used in “inhumane or extremely harmful ways.” Google and Pentagon reportedly agree on deal for ‘any lawful’ use of AI The classified deal apparently doesn’t allow Google to veto how the government will use its AI models. The classified deal apparently doesn’t allow Google to veto how the government will use its AI models. If the agreement is confirmed, it would place Google alongside OpenAI and xAI, which have also made classified AI deals with the US government. Anthropic was also among that list until it…

Musk and Altman go to court

The trial between Elon Musk and OpenAI is officially upon us. And it is going to be a mess. As the two sides fight over the early days of AI, who deserves credit and cash for what, and more, we’re likely to spend the next few weeks hearing a lot of important people’s secrets made extremely public. Which may be exactly what Musk is going for. Musk and Altman go to court On The Vergecast: Musk v. OpenAI, Framework’s new gear, and fast chips for small laptops. On The Vergecast: Musk v. OpenAI, Framework’s new gear, and fast chips for small laptops. Verge subscribers, don’t forget you get exclusive access to ad-free Vergecast wherever you get your podcasts. Head here. Not a subscriber? You can sign up here. On this episode of The Vergecast, The Verge’s Liz Lopatto joins the…

Claude can now plug directly into Photoshop, Blender, and Ableton

Anthropic has launched a set of connectors for Claude that allow the AI chatbot to tap into popular creative software, including Adobe’s Creative Cloud apps, Affinity, Blender, Ableton, Autodesk, and more. Claude can now plug directly into Photoshop, Blender, and Ableton Anthropic is also giving the Blender Foundation a load of cash to help the software stay free and open-source. Anthropic is also giving the Blender Foundation a load of cash to help the software stay free and open-source. This marks the company’s latest efforts to break into the creative industry following its launch of Claude Design earlier this month. The new connectors — which enable Claude to access apps, retrieve data, and take actions within connected services — are “designed to make it easier to use Claude for creative work,” according to Anthropic, and can be used for specific…

Elon Musk takes the stand in high-profile trial against OpenAI

Elon Musk officially began his testimony in the trial he has brought against OpenAI CEO Sam Altman and company president Greg Brockman. Elon Musk takes the stand in high-profile trial against OpenAI The first witness in Musk v. Altman, is Elon Musk himself. The first witness in Musk v. Altman, is Elon Musk himself. The three were on the initial founding team of OpenAI, with Musk investing up to $38 million early on before the co-founders’ relationship soured over disagreements over company structure and mission, including whether or not OpenAI should be folded into Musk-owned Tesla. Musk walked away and, years later, founded xAI — his own direct competitor to OpenAI, which is now owned by Musk’s SpaceX. In recent years, Musk has filed no less than four different lawsuits against OpenAI, many of which have since been dropped or…

Taylor Swift is stepping up the legal war on AI copycats

Taylor Swift has been at the center of AI imitation controversies for years, and now, she’s become the latest celebrity who’s escalating attempts to protect herself from AI copycats. As usual, however, the legal system intersects with technology in complicated ways — and Swift’s efforts may be a long shot. Taylor Swift is stepping up the legal war on AI copycats Trademark applications filed by Taylor Swift could serve as another legal tool to combat AI fakes. Trademark applications filed by Taylor Swift could serve as another legal tool to combat AI fakes. In trademark applications filed last week, Swift’s team asked for protection for two phrases spoken by the singer: Hey, it’s Taylor Swift and Hey, it’s Taylor. The trademark applications, filed by TAS Rights Management on behalf of Swift, include audio clips of Swift saying the two phrases…

Elon Musk tells the jury that all he wants to do is save humanity

On the stand, Elon Musk is positioning himself as a savior. Elon Musk tells the jury that all he wants to do is save humanity His testimony attempted set him up as a magnanimous hero, in contrast to the defendant, Sam Altman. His testimony attempted set him up as a magnanimous hero, in contrast to the defendant, Sam Altman. In the high-profile trial between him and his fellow OpenAI co-founder, now CEO, Sam Altman, Musk opened by going through his background. He went as far back as being raised in South Africa and arriving in Canada for college with “2,500 in Canadian travelers’ checks and a bag of clothes and books,” then spent an unusually long time talking about his past, from Zip2 to PayPal to the current, more familiar slate of companies he now runs. Why is Musk giving…

Elon Musk appeared more petty than prepared

Today the first witness was sworn in in Musk v. Altman: Elon Musk. I was surprised by how flat he seemed. Elon Musk appeared more petty than prepared In his opening testimony against Sam Altman, Musk was unfocused and uncharming. Elon Musk appeared more petty than prepared In his opening testimony against Sam Altman, Musk was unfocused and uncharming. This is not the first time I’ve seen Musk in court. During his defamation suit, he turned on the charm and the jury responded by finding him not guilty. Today he looked adrift and unprepared. The only times he showed real animation were when he was bragging about how much he’d done for OpenAI. The direct examination is a way of telling a story through questions; it’s important to make the narrative clear. For a suit that accuses Sam Altman of…

Run Highly Efficient Multimodal Agentic AI with NVIDIA Nemotron 3 Nano Omni Using vLLM Apr 28, 2026 · 7 min read We are excited to support the newly released NVIDIA Nemotron 3 Nano Omni model on vLLM.

Run Highly Efficient Multimodal Agentic AI with NVIDIA Nemotron 3 Nano Omni Using vLLM We are excited to support the newly released NVIDIA Nemotron 3 Nano Omni model on vLLM. Nemotron 3 Nano Omni, part of the Nemotron 3 family of open models, is the highest efficiency, open multimodal model with leading accuracy, built to power sub-agents that perceive and reason across vision, audio, and language in a single loop. Enterprise agent workflows are inherently multimodal. Agents must interpret screens, documents, audio, video, and text, often within the same reasoning pass. Yet most agentic systems today bolt together separate models for vision, speech, and language, multiplying inference hops, complicating orchestration, and fragmenting context across the pipeline. Nemotron 3 Nano Omni addresses two major challenges this fragmentation creates: - Fragmented Models: Running separate vision, audio, and language models in sequence increases…

These Men Allegedly Profit Off Teaching People How to Make AI Porn

A little over a year ago, MG was leading the relatively normal life of a twentysomething in Scottsdale, Arizona. She worked as a personal assistant and supplemented her income by waiting tables on the weekends. Like most women her age, she had an Instagram account, where she’d occasionally post Stories and photos of herself getting matcha and hanging out by the pool with her friends, or going to Pilates. “I never really cared to pop off and become popular on social media,” says MG (who is cited only as MG in the lawsuit to protect her identity). “I just used it the way most people did when it first came out, to share their lives with the people closest to them.” She has a little more than 9,000 followers—a robust following, but nowhere close to a massive platform. Last summer,…

OpenAI Rolls Out ‘Advanced’ Security Mode for At-Risk Accounts

For anyone who fears their ChatGPT and Codex accounts might be targeted by attackers, OpenAI announced on Thursday that it is adding an optional new level of account protection that adds an extra layer of security. Dubbed Advanced Account Security, the feature enforces strict access controls that would make account takeover attacks very difficult. Such measures are not a new idea in the realm of account security. Google, for example, has offered its Advanced Protection account security tier for nearly a decade. But as mainstream AI services rapidly proliferate around the world, there is a pressing need for an array of basic protections to be put in place. OpenAI says the launch is part of its broader cybersecurity strategy announced earlier this month. “People are turning to AI for deeply personal questions and increasingly high-stakes work,” the company said on…

Elon Musk Seemingly Admits xAI Has Used OpenAI’s Models to Train Its Own

While testifying on Thursday in federal court, Elon Musk seemed to indicate that his AI lab may have used OpenAI’s models to train xAI’s own. He touched upon the topic while sitting on the witness stand answering cross-examination questions from an OpenAI attorney amid his ongoing legal battle against the ChatGPT-maker. This is the exchange, as best as WIRED could capture it: OpenAI Lawyer William Savitt: Do you know what distillation is? Musk: It means to use one AI model to train another AI model. Savitt: Has xAI done that with OpenAI? Musk: Generally all the AI companies [do that]. Savitt: So that’s a yes. Musk: Partly. Distillation is a technique where a smaller AI model is trained to mimic the behavior of a larger, more capable model, making it cheaper and faster to run while preserving much of its…

Musk v. Altman Kicks Off, DOJ Guts Voting Rights Unit, and Is the AI Job Apocalypse Overhyped?

This week on Uncanny Valley, the team discusses the stakes behind the trial of Elon Musk against OpenAI’s leadership (and how Microsoft is trying to stay away from the drama). They also look into what recent layoffs announced at Meta and the industry at large say about the ways in which AI is—and isn’t—replacing jobs. Also, we dive into a WIRED investigation on how the Department of Justice has effectively hollowed out its voting rights work, and how this move could impact future elections. Articles mentioned in this episode: - Musk v. Altman Is a Battle for OpenAI’s Soul - Some Musk v. Altman Jurors Don't Like Elon Musk - ‘It’s Undignified’: Hundreds of Workers Training Meta’s AI Could Be Laid Off - ‘The Damage Is Massive’: How the Justice Department Dismantled Its Voting Rights Section You can follow Brian…

Good Luck Getting a Mac Mini for the Next ‘Several Months’

Apple CEO Tim Cook said on the company’s earnings call on Thursday that it could take “several months” to meet skyrocketing demand for the Mac Mini, the company’s compact but mighty, screen-free desktop computer. Cook’s remarks come after coders determined in recent months that the Mac Mini was the perfect machine for agentic AI tasks. “On the Mac Mini and Mac Studio, both of these are amazing platforms for AI and agentic tools,” Cook said on the earnings call, in response to analyst questions. “And customer adoption of that is happening faster than we expected.” The news comes amid another record-setting quarter for the company. iPhone sales came up shorter than expected, though demand for the iPhone 17 has been super high, and Apple’s subscription services business has continued to grow. Apple faced supply constraints on both the iPhone and…

How AI Could Help Combat Antibiotic Resistance

Antibiotic resistance is a fast-growing public health crisis, causing more than a million global deaths annually and contributing to nearly 5 million more. These infections are more difficult and more expensive to treat than typical infections and are responsible for longer hospital stays, driving up costs for hospitals and patients alike. Treatment mostly comes down to guesswork on the part of physicians. Ara Darzi, a surgeon and director of the Institute of Global Health Innovation at Imperial College London, says AI-powered diagnostics offer a better way. “We're standing, right now, in 2026, at the first genuine inflection point in this crisis,” Darzi said on April 16 at WIRED Health in London. The overuse and misuse of antibiotics and a lack of new drug development have been fueling the rise of resistant microbes. When bacteria are exposed to levels of antibiotics…

I've Covered Robots for Years. This One Is Different

A robot’s claw hurtles toward a light bulb on a table. I wince, waiting for the crunch. But suddenly the claw decelerates. It starts gingerly pawing around the table, as if searching for its glasses on the nightstand. It gently positions the bulb between its two pincers. The bulb rolls away. The claw goes chasing it across the table. After a few nips, the bulb is back in its grasp. The robot swiftly screws the bulb into a nearby socket, illuminating its work area. In more than a decade of writing about robots, I have never seen one move so naturally. Most are ham-fisted klutzes, even when remotely controlled by a person. Of the few dozen robot arms on the market today, not one can screw in a light bulb. I have come to visit Eka, a startup located in…

Sanctioned Chinese AI Firm SenseTime Releases Image Model Built for Speed

SenseTime, a Chinese AI company best known for its facial recognition technology, released a new open source model on Tuesday that it claims can both generate and interpret images far faster than top models developed by US competitors. SenseNova U1 could help the company reclaim lost ground after it slipped from its place among the leading players in China’s AI development race. The model’s secret sauce is its ability to “read” images without translating them to text first, speeding up the process and reducing the amount of computing power required. “The model’s entire reasoning process is no longer limited to text. It can reason with images as well,” Dahua Lin, cofounder and chief scientist at SenseTime, said in an interview with WIRED. Lin, who is also a professor of information engineering at the Chinese University of Hong Kong, says that…

Taylor Swift Wants to Trademark Her Likeness. These TikTok Deepfake Ads Show Why

Last week, Taylor Swift filed a trio of trademark applications to protect her image and voice. One is meant to cover a well-known photograph of the pop singer holding a pink guitar during a concert on her record-breaking Eras tour, while the two sound trademarks are for simple identifying phrases: “Hey, it’s Taylor Swift” and “Hey, it’s Taylor.” The move comes as AI deepfakes continue to proliferate across social media. Any individual stands to have their likeness exploited in the creation of nonconsensual AI-generated material; earlier this month, an Ohio man was the first person convicted under a new federal law criminalizing “intimate” visual deceptions of this sort. Celebrities, meanwhile, find themselves at risk of both explicit deepfakes and false endorsements. A new report from the AI detection company Copyleaks shows that Swift and other stars have recently had their…

Emergency First Responders Say Waymos Are Getting Worse

Emergency first-responder leaders told federal regulators in a private meeting last month that they were frustrated with the performance of autonomous vehicles on their streets—that city firefighters, police officers, EMTs, and paramedics are forced to spend time during emergencies resolving issues with frozen or stuck cars. One fire official called them “a safety issue for our crews as well as the victims.” WIRED obtained an audio recording of the meeting. Officials from San Francisco and Austin, where Waymo has been ferrying passengers without drivers for more than a year, said the vehicles’ performance is getting worse. “We are actually seeing something interesting: backsliding of some things that had improved upon,” Mary Ellen Carroll, the executive director of San Francisco’s Department of Emergency Management, told officials with the National Highway Traffic Safety Administration (NHTSA), which oversees self-driving vehicle safety in the…

How Elon Musk Squeezed OpenAI: They 'Are Gonna Want to Kill Me’

Elon Musk returned to the witness stand on Wednesday to continue telling his side of the story in his legal battle against OpenAI and its CEO Sam Altman. Under cross-examination from OpenAI’s lawyers, Musk was pressed on all the ways he tried to squeeze the organization over a 2017 power struggle that he ultimately lost. Around this time, Musk tried to hire away OpenAI researchers and stopped sending it funding he had previously promised, according to emails presented as evidence in the case. As the cross-examination began, tension rippled through the courtroom. Judge Yvonne Gonzalez Rogers started the day by reprimanding someone in the gallery for taking a picture of Musk. OpenAI’s president and cofounder, Greg Brockman, sat behind his lawyers with a yellow legal pad in his lap, giving Musk a cold stare as he testified. Musk grew visibly…

The Bloomberg Terminal Is Getting an AI Makeover, Like It or Not

For its famous intractability, the Bloomberg Terminal has long inspired devotion, bordering on obsession. Among traders, the ability to chart a path through the software’s dizzying scrolls of numbers and text to isolate far-flung information is the mark of a seasoned professional. But as a greater mass of data is fed into the Terminal—not only earnings and asset prices, but weather forecasts, shipping logs, factory locations, consumer spending patterns, private loans, and so on—valuable information is being lost. “It has become more and more untenable,” says Shawn Edwards, chief technology officer at Bloomberg. “You miss things, or it takes too long.” To try to remedy the problem, Bloomberg is testing a chatbot-style interface for the Terminal, ASKB (pronounced ask-bee), built atop a basket of different language models. The broad idea is to help finance professionals to condense labor-intensive tasks, and…

The Race Is on to Keep AI Agents From Running Wild With Your Credit Cards

Between malware, online impersonation, and account takeovers, there are enough digital security problems out there as it is. And with the rise of agentic AI, more activity is being carried out by agents on behalf of humans—creating different risks that something could go awry. Now, working with initial contributions from Google and Mastercard, the authentication-focused industry association known as the FIDO Alliance said on Tuesday that it will launch a pair of working groups to develop industry standards for validating and protecting payments and other transactions carried out by AI agents. The goal is to produce a protective baseline that can be adopted across industries. This way, users can authorize agent actions using mechanisms that can't easily be phished, or taken over by a bad actor to give an agent rogue instructions. The standards would also include cryptographic tools that…

‘It’s Undignified’: Hundreds of Workers Training Meta’s AI Could Be Laid Off

Hundreds of workers in Ireland tasked with refining Meta’s AI models have been told that their jobs are at risk as the company embarks on a sweeping new round of layoffs, according to documents obtained by WIRED. The affected workers are employed by the Dublin-based firm Covalen, which handles various content moderation and labeling services for Meta. The workers were informed of the layoffs over a brief video meeting on Monday afternoon and were not allowed to ask questions, according to Nick Bennett, one of the employees on the call. “We had a pretty bad feeling [before the meeting],” he says. “This has happened before.” In all, more than 700 employees stand to potentially lose their jobs at Covalen, according to an email reviewed by WIRED. Roughly 500 are data annotators. Their job is to check material generated by Meta’s…

Elon Musk Testifies That He Started OpenAI to Prevent a ‘Terminator Outcome’

Elon Musk and Sam Altman appeared in a federal courtroom together for the first time on Tuesday as they fight over OpenAI’s decade-long evolution and what it means for the company’s future. The trial in Musk’s lawsuit against Altman could result in financial damages and, more significantly, governance changes at OpenAI that may complicate its plans for an initial public offering as soon as this year. As the first witness on the stand, Musk immediately sought to frame his case as more than just about OpenAI. Siding with Altman “will give license to looting every charity in America” and shake the “entire foundation of charitable giving,” Musk told a panel of nine jurors advising US District Judge Yvonne Gonzalez Rogers on how to rule. Musk has been concerned about computers becoming smarter than people “since he was a young man…

OpenAI Really Wants Codex to Shut Up About Goblins

OpenAI has a goblin problem. Instructions designed to guide the behavior of the company’s latest model as it writes code have been revealed to include a line, repeated several times, that specifically forbids it from randomly mentioning an assortment of mythical and real creatures. “Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user’s query,” read instructions in Codex CLI, a command-line tool for using AI to generate code. It is unclear why OpenAI felt compelled to spell this out for Codex—or indeed why its models might want to discuss goblins or pigeons in the first place. The company did not immediately respond to a request for comment. OpenAI’s newest model, GPT-5.5, was released with enhanced coding skills earlier this month. The company is in a…