Raw English sources. Zero distortion.

No hype. No fluff. No ads.

May 3, 2026
Selected from 43 source items
  1. 01 Microsoft is adding a workflow-first Legal Agent to Word to help legal teams review and negotiate documents.
  2. 02 Meta has acquired Assured Robot Intelligence to strengthen its humanoid-robotics model work.
  3. 03 A new test compares AI-powered dictation apps for email, note-taking and even coding by voice.
May 2, 2026
Selected from 64 source items
  1. 01 The Pentagon added OpenAI, Google, Microsoft, Nvidia, Amazon and others to classified AI contracts — notably excluding Anthropic.
  2. 02 Elon Musk’s trial testimony against OpenAI continued to surface internal messages and tensions over the company’s structure.
  3. 03 OpenAI published a new Advanced Account Security program with phishing‑resistant login and stronger recovery protections.
May 1, 2026
Selected from 69 source items
  1. 01 Elon Musk testified that xAI used OpenAI’s models to train its Grok assistant, putting distillation at the center of the Musk‑Altman trial.
  2. 02 Meta is promoting acquisition Manus’s AI with ads promising quick business income even as the company touts 10 million weekly business conversations.
  3. 03 OpenAI published a technical post tracing the origins and fixes for “goblin” outputs in GPT‑5 and acknowledged the behavior as a learned quirk.
April 30, 2026
Selected from 67 source items
  1. 01 Google Photos can create a virtual wardrobe from your gallery, letting you mix, save and share outfit looks.
  2. 02 OpenAI outlined a five-part plan to bolster cybersecurity and broaden AI-assisted cyber defense.
  3. 03 Google marks 20 years of Translate with tips, feature highlights and anniversary notes.
April 29, 2026
Selected from 69 source items
  1. 01 Reporting says Google signed a classified deal allowing the Department of Defense to use its AI models for “any lawful government purpose.”
  2. 02 OpenAI secured FedRAMP Moderate authorization for ChatGPT Enterprise and the OpenAI API, clearing a path to federal use.
  3. 03 DARPA-backed teams scanned 54 million lines of injected code to stress-test AI vulnerability-detection systems.
April 28, 2026
Selected from 50 source items
  1. 01 OpenAI and Microsoft agreed an amended partnership that ends exclusivity and lets OpenAI sell on other clouds, including Amazon Bedrock.
  2. 02 Jury selection began in Elon Musk’s suit against OpenAI and Sam Altman over the company's governance and mission.
  3. 03 More than 600 Google employees urged CEO Sundar Pichai to block classified Pentagon use of Google’s AI models.
April 27, 2026
Selected from 29 source items
  1. 01 Anthropic ran an agent‑on‑agent marketplace that executed real transactions, including a Mill Valley property offered in exchange for company equity.
  2. 02 AI agents in the test represented buyers and sellers and completed settlements for real goods and money.
  3. 03 Agent marketplaces could scale commerce in new ways while raising oversight, billing and legal responsibility questions.
April 26, 2026
Selected from 53 source items
  1. 01 Anthropic ran a classified marketplace experiment where autonomous agents represented buyers and sellers and completed real transactions.
  2. 02 Claude gained connectors to personal services including Spotify, Uber Eats, Instacart and TurboTax.
  3. 03 China’s DeepSeek released a V4 model preview that it says can compete with major US systems, especially on coding tasks.
April 25, 2026
Selected from 75 source items
  1. 01 Google pledges up to $40 billion in cash and compute to Anthropic to secure large‑model capacity.
  2. 02 OpenAI unveiled GPT‑5.5, positioning it as faster and more capable for coding, research, and data work.
  3. 03 Tim Cook will step down in September, handing the CEO role to hardware chief John Ternus as the App Store landscape evolves.
April 24, 2026
Selected from 82 source items
  1. 01 OpenAI introduced cloud-based workspace agents in ChatGPT to automate repeatable team workflows and connect tools across Business, Enterprise, Edu, and Teachers plans.
  2. 02 OpenAI opened a GPT-5.5 Bio Bug Bounty offering rewards up to $25,000 to find jailbreaks and bio-safety risks.
  3. 03 The White House warned of “industrial‑scale” efforts in China to rip off U.S. AI technology, highlighting national-security concerns around AI supply chains.
April 23, 2026
Selected from 76 source items
  1. 01 SpaceX struck a deal to partner with Cursor and secured an option to buy the AI coding startup for $60 billion — or pay a $10 billion fee if it walks away.
  2. 02 Google unveiled two eighth‑generation TPUs, splitting training and inference tasks to better serve agentic AI workloads.
  3. 03 OpenAI is offering ChatGPT for Clinicians at no cost to verified U.S. physicians, nurse practitioners, and pharmacists to aid care and documentation.
April 22, 2026
Selected from 65 source items
  1. 01 Ars Technica reports Meta will use employee mouse and keyboard tracking to collect interactive data for training AI agents.
  2. 02 OpenAI announces Codex Labs and enterprise partnerships as Codex reaches 4 million weekly active users.
  3. 03 The Pentagon seeks a $54B drone investment while Anthropic’s Mythos raises cyber‑security and procurement questions.
April 21, 2026
Selected from 43 source items
  1. 01 A developer built Explain The Law using Gemini 2.5 Pro to generate anchored, statically generated summaries of bills and executive orders.
  2. 02 Reports say the NSA is using Anthropic’s restricted Mythos model even as researchers warn the model could accelerate exploitation of cyberdefenses.
  3. 03 Hyatt has deployed ChatGPT Enterprise across its global workforce, citing GPT‑5.4 and Codex to boost productivity and guest operations.
April 19, 2026
Selected from 47 source items
  1. 01 Anthropic’s Mythos Preview — a cybersecurity-focused model — appears to have eased tensions with the Trump administration after months of public conflict and a Pentagon supply‑chain designation.
  2. 02 OpenAI is winding down Sora and folding its science team as leaders including Bill Peebles and Kevin Weil depart the company.
  3. 03 Talk of “tokenmaxxing” and a wave of OpenAI deals underscore a growing gap between AI insiders’ priorities and public scrutiny.
April 18, 2026
Selected from 73 source items
  1. 01 OpenAI shut down Sora and the team lead, Bill Peebles, has left as the company refocuses on enterprise AI.
  2. 02 OpenAI updated Codex for macOS and Windows with in-app browsing, image generation, memory, and plugins.
  3. 03 Anthropic released a cybersecurity-focused Mythos Preview that may soften government scrutiny around the company.
April 17, 2026
Selected from 77 source items
  1. 01 OpenAI launches GPT‑Rosalind, a reasoning model positioned to accelerate drug discovery, protein reasoning, and genomics workflows.
  2. 02 Anthropic ships Claude Opus 4.7 as its most capable generally available Opus release for complex coding and image analysis.
  3. 03 Google expands Gemini’s personalization with Photos integration for tailored image generation in the Gemini app.
April 16, 2026
Selected from 88 source items
  1. 01 OpenAI’s Codex gains desktop control, image generation and memory in a clear bid to match Anthropic’s developer tooling.
  2. 02 Anthropic ships Claude Opus 4.7 as a generally available model tuned for complex engineering work.
  3. 03 Google’s Gemini can now draw on Google Photos to create images tailored to a user’s personal context.
April 15, 2026
Selected from 37 source items
  1. 01 OpenAI adds sandboxed, model-native execution to its Agents SDK to help enterprises run longer, safer agent workflows.
  2. 02 Allbirds will sell its brand and assets and reposition toward AI services, a move that sent its stock sharply higher.
  3. 03 Google expands Gemini to the Mac and adds expressive Flash TTS as it rolls models into desktop and robotics use cases.
April 14, 2026
Selected from 20 source items
  1. 01 Ukraine is deploying more robots into frontline kill zones to reduce soldiers’ exposure to drone hazards.
  2. 02 Chrome’s new Skills feature turns saved Gemini prompts into one‑click tools for common workflows.
  3. 03 The UK’s Mythos model completed a multistep infiltration challenge meant to probe AI cybersecurity risk.
April 13, 2026
Selected from 10 source items
  1. 01 Meta is training an AI avatar of Mark Zuckerberg to converse with employees, with the CEO personally involved in its development.
  2. 02 College instructors say widespread LLM use has become the most demoralizing challenge of teaching today.
  3. 03 Cloudflare’s Agent Cloud runs OpenAI’s GPT-5.4 and Codex to help enterprises build and scale production AI agents.
April 12, 2026
Selected from 32 source items
  1. 01 OpenAI released a plain‑English primer explaining what AI is and how ChatGPT uses large language models.
  2. 02 The company says it rotated macOS code‑signing certificates, pushed app updates, and found no user data was exposed after the Axios developer‑tool supply‑chain compromise.
  3. 03 OpenAI also published practical safety guidance with concrete best practices for accuracy, transparency, and responsible use.
April 11, 2026
Selected from 82 source items
  1. 01 OpenAI published step‑by‑step guides showing how to use ChatGPT to gather sources, analyze findings, and produce citation‑backed research.
  2. 02 CyberAgent says ChatGPT Enterprise and Codex are accelerating decisions across advertising, media, and gaming.
  3. 03 OpenAI also rolled out academy lessons for operations, sales, and customer‑success teams to standardize workflows and outreach with ChatGPT.
April 10, 2026
Selected from 68 source items
  1. 01 Meta’s Muse Spark model pushed the Meta AI app from about No. 57 to No. 5 on the App Store after it began powering the app and web experience.
  2. 02 OpenAI released a Child Safety Blueprint that lays out safeguards and age‑appropriate design to address harms including child sexual exploitation.
  3. 03 OpenAI published a roadmap for enterprise adoption that emphasizes Frontier, ChatGPT Enterprise, Codex, and company‑wide AI agents.
April 9, 2026
Selected from 69 source items
  1. 01 YouTube Shorts launched an AI avatar tool that can generate realistic creator clones for video.
  2. 02 Anthropic previewed Mythos and Project Glasswing, pitching AI to surface vulnerabilities for large enterprises.
  3. 03 OpenAI published a Child Safety Blueprint proposing age‑appropriate safeguards and steps to counter AI‑linked online exploitation.
April 8, 2026
Selected from 64 source items
  1. 01 OpenAI opens a pilot Safety Fellowship to fund independent alignment research and grow next-gen safety talent.
  2. 02 Anthropic’s Project Glasswing will let major cloud and tech firms use its model to surface vulnerabilities across operating systems and browsers.
  3. 03 Reporting and previews say Anthropic’s newest model, Mythos, is powerful enough that the company limited its release and used it only with select partners for defensive work.
April 7, 2026
Selected from 47 source items
  1. 01 Intel is betting advanced chip packaging will help it capture more of the AI hardware market.
  2. 02 Google quietly shipped an iOS dictation app that runs Gemma models offline.
  3. 03 OpenAI opened a pilot Safety Fellowship to fund independent alignment research.
April 3, 2026
Selected from 60 source items
  1. 01 The Claude Code leak exposes Anthropic’s work on a persistent agent, a stealth ‘Undercover’ mode, and an apparent assistant called Buddy — then a DMCA sweep took down legitimate GitHub forks.
  2. 02 OpenAI’s Gradient Labs will use GPT‑4.1 and GPT‑5.4 mini/nano agents to automate bank account management for customers.
  3. 03 Google added Flex and Priority inference dials to the Gemini API so developers can trade cost against latency and reliability.
April 2, 2026
Selected from 60 source items
  1. 01 Leaked Claude Code reveals an always-on agent, a stealth “Undercover” mode and a Tamagotchi-like virtual pet buried in a released package.
  2. 02 Alexa+ now supports conversational ordering from Grubhub and Uber Eats, letting users build and modify orders like they would with a waiter.
  3. 03 OpenAI’s Gradient Labs announces AI account managers for banks powered by GPT-4.1 and GPT-5.4 mini/nano models.
March 31, 2026
Selected from 42 source items
  1. 01 OpenAI abruptly closed Sora six months after launch, prompting scrutiny over user-face uploads and what it means for AI-generated video products.
  2. 02 Bluesky’s new Attie app, built on Anthropic’s Claude, gives users tools to design and control their own feed algorithm.
  3. 03 Import AI examines the idea of 'political superintelligence' alongside Google’s multi-agent robotics work and a robot drummer.
March 30, 2026
Selected from 31 source items
  1. 01 Bluesky released Attie, an AI assistant that helps users assemble personalized feeds on the AT Protocol.
  2. 02 Attie recommends accounts, topics, and filters so a feed reflects intent instead of raw follows.
  3. 03 The release underscores Bluesky’s move toward developer-facing AI tools for customizable open social networking.
March 29, 2026
Selected from 45 source items
  1. 01 Developers on Hacker News describe simple observability and per-agent limits that stop coding agents from running up large model bills.
  2. 02 A thin proxy that tags requests with agent/task context can reveal which agent is costly and let teams block it.
  3. 03 Legal and procurement drama — like a judge’s injunction for Anthropic and turnover in the White House AI advisor role — is increasing vendor uncertainty for government and enterprise users.
March 28, 2026
Selected from 68 source items
  1. 01 Anthropic says paid Claude subscriptions have more than doubled this year, even as outside estimates put total consumer users between 18 million and 30 million.
  2. 02 All but two of xAI’s original 11 co‑founders had departed before this week, and the last reported co‑founder has now left.
  3. 03 A judge granted Anthropic a preliminary injunction that temporarily blocks the Pentagon’s ban while litigation proceeds; the case follows weeks of dispute over a supply‑chain designation.
March 26, 2026
Selected from 30 source items
  1. 01 OpenAI's newly announced Foundation will invest at least $1 billion across health, economic opportunity, AI resilience, and community programs.
  2. 02 The administration's President’s Council of Advisors on Science and Technology will include top executives such as Mark Zuckerberg and Jensen Huang, even as Congress moves to codify safety limits around military AI.
  3. 03 AI startups continue to attract funding: Granola jumps to a $1.5B valuation, Mirage and others raise growth rounds, and privacy-focused rivals emphasize local-first designs.
March 25, 2026
Selected from 34 source items
  1. 01 OpenAI announced it is winding down Sora and its social video app despite a recent Disney licensing tie-up.
  2. 02 Anthropic launched an ‘auto mode’ for Claude Code to let models act with limited autonomy while keeping safety controls.
  3. 03 Sift’s founders are porting the data and infrastructure used at SpaceX to help factories scale advanced manufacturing.
March 24, 2026
Selected from 24 source items
  1. 01 ChatGPT and Gemini are adding richer shopping features and merchant integrations, even as OpenAI pulls back on some instant-checkout plans.
  2. 02 Anthropic’s Claude Code and Cowork now perform actions on users’ computers with reduced confirmation in an ‘auto’ mode that preserves built-in safeguards.
  3. 03 OpenAI released prompt-based teen safety policies and open-source tools to help developers moderate age-specific risks in AI experiences.

View full archive →

Subscribe to AI Digest

A compact morning briefing built from primary AI sources.