May 2026

Sam Altman: OpenAI launches robotics division, hiring engineers to build physical AI

Sam Altman announces OpenAI Robotics, evolved from world simulation research, and opens hiring for hardware and ML engineers to build robots for skilled workers and eventually personal use.

Sam Altman: OpenAI launches robotics division, hiring engineers

Sam Altman announced OpenAI Robotics is hiring engineers to build useful robots, evolving from the world simulation research program.

Sam Altman: OpenAI launches robotics division, hiring engineers

Altman announces OpenAI Robotics, evolved from world simulation research, and is hiring engineers to build robots for skilled workers and eventually personal use.

Sam Altman: OpenAI launches robotics division to build useful physical-world robots

Altman announces OpenAI Robotics, evolved from world simulation research, is hiring full-stack hardware and ML engineers to build robots for skilled workers and eventually personal use.

Sam Altman: OpenAI Robotics launches, hiring engineers to build useful robots

Sam Altman announces OpenAI Robotics, spun out of world simulation research, and opens hiring for engineers to build robots for skilled workers and eventually personal use.

Sam Altman: OpenAI launches robotics division, hiring engineers to build physical AI

Altman announces OpenAI Robotics, evolved from world simulation research, and is hiring engineers to build robots that assist skilled workers and eventually serve as personal assistants.

Sam Altman: OpenAI debuts Rosalind biodefense initiative

Sam Altman announced OpenAI's new effort to help the world prepare for biological threats through its Rosalind biodefense program.

Anthropic has released Claude Opus 4.8, bringing sharper coding performance, on-the-fly dynamic workflows with coordinated subagents, and a cheaper fast mode to Claude Code and APIs.

Claude: How red teams break Claude before launch to improve it

Claude shares insight into its pre-release process, where internal teams stress-test new models to identify weaknesses and improve quality.

Anthropic: Anthropic run-rate revenue surpasses $47 billion

Anthropic announced its run-rate revenue crossed $47 billion, driven by enterprise and consumer adoption of Claude, alongside its Series H announcement.

Anthropic: Anthropic Raises $65B Series H at $965B Valuation

Anthropic raised $65 billion in Series H funding at a $965 billion post-money valuation led by Altimeter, Dragoneer, Greenoaks, and Sequoia.

Anthropic: Anthropic Raises $65B Series H at $965B Valuation

Anthropic raised $65 billion in Series H funding at a $965 billion post-money valuation led by Altimeter, Dragoneer, Greenoaks, and Sequoia.

Anthropic: Anthropic Raises $65B Series H at $965B Valuation

Anthropic raised $65 billion in Series H funding at a $965 billion post-money valuation led by Altimeter, Dragoneer, Greenoaks, and Sequoia.

Min Choi: Claude Opus 4.8 drops with major coding and agent upgrades

Anthropic released Claude Opus 4.8 featuring improved coding, enhanced agent capabilities, effort control, and dynamic workflows with hundreds of subagents, plus Mythos-class models coming soon.

Claude: Anthropic launches Claude Opus 4.8 with sharper judgment

Claude announced Opus 4.8, featuring sharper judgment, improved honesty about its progress, and longer independent work at the same price.

Claude: Anthropic launches Claude Opus 4.8 with sharper judgment

Claude Opus 4.8 is now available, featuring improved judgment, greater self-assessment honesty, and extended independent operation at unchanged pricing.

Claude: Anthropic launches Claude Opus 4.8 with sharper judgment

Claude Opus 4.8 launches with improved judgment, greater honesty about its capabilities, and extended independent work sessions at unchanged pricing.

Stability AI has released Stable Audio 3.0 as open-weight models trained on fully licensed data, introduced Brand Studio for enterprise creative production, and formed partnerships with Warner Music Group, Universal Music Group, EA, NVIDIA, and Amazon Web Services.

Replicate's blog lists recent posts covering video and image generation models such as Seedance 2.0, Seedream 5.0, and FLUX.2, alongside prompting guides and company news including Replicate joining Cloudflare. Source: Replicate (https://replicate.com/blog).

Mistral AI announced multiple product and research updates, including remote coding agents in Vibe powered by Mistral Medium 3.5, the Forge system for enterprise AI models, Voxtral TTS, and a partnership with NVIDIA. Other releases include Mistral Small 4, Leanstral for Lean 4, Workflows in public preview, and MCP connectors in Studio.

Hugging Face's blog listing showcases recent community and official articles covering open-source humanoid robotics, audio LLMs, transformer inference optimization, and agent benchmarks. The posts span topics from fine-tuning video generation models to local browser deployment and reinforcement learning post-training techniques.

Google DeepMind's blog features multiple announcements including the Gemini Omni and Gemini 3.5 models, a new national AI partnership with Singapore, scientific research tools like Co-Scientist and AlphaEvolve, and the Google DeepMind Accelerator program for environmental risks in Asia Pacific.

Anthropic's newsroom highlights recent launches of Claude Opus 4.7 and Claude Design, the Project Glasswing security initiative, and new enterprise alliances with KPMG, PwC, and the Gates Foundation.

OpenAI's blog lists recent posts from May 2026 spanning self-improving tax agents with Codex, a Gartner leadership recognition for enterprise coding agents, a disproven discrete geometry conjecture by an OpenAI model, content provenance efforts, a Dell Technologies partnership, new ChatGPT personal finance features, and safety improvements for sensitive conversations.

Claude: Replit's Michele Catasta on 50M users building with Claude

Replit President Michele Catasta's mission to democratize software has reached 50 million users building on the platform with Claude.

Claude: Replit's AI chief on building software for 50M users with Claude

Claude spotlights Replit President Michele Catasta and how the platform enables over 50 million people to build software using natural language.

Sam Altman: OpenAI Foundation commits $250M to AI prosperity

Sam Altman announces the OpenAI Foundation's initial $250 million commitment to support measurement, transition, and shared prosperity in the age of AI.

Sam Altman: OpenAI Foundation commits $250M to AI prosperity and measurement

Sam Altman announces the OpenAI Foundation's initial $250 million commitment to measurement, transition support, and new approaches for broadly shared prosperity in the age of AI.

Andrej Karpathy: Geoff Hinton's official Google title was once "intern"

Karpathy recalls that Geoffrey Hinton's official title at Google was once listed as "intern."

Andrej Karpathy: Geoff Hinton's official title at Google was once "intern"

Karpathy recalls that Geoff Hinton's official title at Google was once listed as "intern".

Andrej Karpathy: Geoff Hinton's official Google title was once "intern"

Karpathy shares that Geoff Hinton once held the official title of "intern" at Google.

Andrej Karpathy: Geoff Hinton was once officially an 'intern' at Google

Karpathy shares that Geoff Hinton's official title at Google was once "intern".

Andrej Karpathy: Geoff Hinton once held the official title of 'intern' at Google

Andrej Karpathy shares an anecdote that Geoff Hinton was officially titled an "intern" during his time at Google.

Andrej Karpathy: Karpathy praises big tech firm for cutting through industry BS

Karpathy applauds a large company for pushing back on hype and highlights a focused technical team executing with clarity.

Andrej Karpathy: Refreshing to see a big company successfully call out industry bs

Karpathy praises a large company for successfully pushing back against industry nonsense with a focused team.

Andrej Karpathy: Praising a big company for calling out industry BS

Karpathy praises a large company for pushing back against widespread nonsense, highlighting a focused team of MTS engineers on a mission.

Andrej Karpathy: Praising a big company for calling out industry BS

Karpathy praises a large company for bluntly rejecting nonsense and highlights a focused MTS team on a mission.

Demis Hassabis: Hassabis praises Garry Tan's Gemini Live use

Demis Hassabis replies to Garry Tan praising his use of Gemini Live and the work he's doing with it.

Demis Hassabis: Hassabis cheers Gemini Live success with Garry Tan

Demis Hassabis replies to Garry Tan about Gemini Live performing well, expressing excitement about the use cases Tan is building with the product.

Demis Hassabis: Hassabis congratulates Cleo Abram on her amazing work

Demis Hassabis sends congratulations to Cleo Abram and encourages her to continue her amazing work.

Demis Hassabis: Congrats to Cleo Abram on her amazing work

Demis Hassabis congratulates Cleo Abram and encourages her to keep up the amazing work.

Demis Hassabis: Enjoyed chatting with Two Minute Papers

Demis Hassabis shares that he had fun chatting with the Two Minute Papers channel.

Sam Altman: Altman asks what problems users want AI to solve

Sam Altman crowdsources ideas by asking followers which problems they most hope artificial intelligence will solve in the future.

Sam Altman: Altman asks followers which problems they want AI to solve

Sam Altman crowdsources ideas from the community, asking what problems people most hope AI will solve and suggesting OpenAI may be able to help.

Sam Altman: Altman asks what problems users want AI to solve

Sam Altman crowdsources ideas by asking followers which problems they most hope AI will solve in the future.

Sam Altman: OpenAI launches new Codex today

Sam Altman announces that a new version of Codex, OpenAI's coding agent, is shipping today.

Sam Altman: Altman condemns chilling San Diego mosque attack

Sam Altman expresses condolences and describes the attack at a San Diego mosque as one of the most chilling incidents he has witnessed.

Demis Hassabis: Hassabis thanks NBTJacklyn for enjoyable conversation

Demis Hassabis thanks NBTJacklyn and says he really enjoyed their conversation.

Demis Hassabis: Thanks NBTJacklyn for the fun conversation

Demis Hassabis thanks NBTJacklyn for a super fun chat and says he really enjoyed the conversation.

Demis Hassabis: Hassabis thanks NBTJacklyn for enjoyable conversation

Demis Hassabis thanks NBTJacklyn and says he really enjoyed their conversation.

Demis Hassabis: More info on Gemini 3.5 Flash model released

Demis Hassabis shares a link to Google's blog post with more information about the Gemini 3.5 Flash model.

Demis Hassabis: Hassabis shares details on new Gemini 3.5 Flash model

Demis Hassabis shares a link to Google's blog post with more information on the Gemini 3.5 Flash model.

Demis Hassabis: Gemini 3.5 Flash outperforms 3.1 Pro with 4x speed boost

Demis Hassabis announces Gemini 3.5 Flash, highlighting superior coding performance, 4x faster speed than frontier models, and lower cost, with Pro version coming soon.

Demis Hassabis: Gemini 3.5 Flash launches with 4x speed and half the cost

Demis Hassabis introduces Gemini 3.5 Flash, highlighting superior coding and agentic performance over 3.1 Pro, 4x speed gains, and lower costs, now available in Antigravity and Gemini App.

Demis Hassabis: Gemini 3.5 Flash outperforms 3.1 Pro, runs 4x faster, costs less

Hassabis introduces Gemini 3.5 Flash, highlighting superior coding performance, 4x speed over frontier models, and lower cost.

Demis Hassabis: Gemini 3.5 Flash outperforms 3.1 Pro at lower cost

Demis Hassabis announces Gemini 3.5 Flash, highlighting superior coding performance, 4x speed, and half the cost of frontier models.

Demis Hassabis: Gemini 3.5 Flash outperforms 3.1 Pro with 4x speed, half cost

Google launches Gemini 3.5 Flash, a model that beats 3.1 Pro on coding and agentic tasks while running 4x faster and often costing less than half the price.

Demis Hassabis: Gemini 3.5 Flash outperforms 3.1 Pro, 4x faster, cheaper

Hassabis touts Gemini 3.5 Flash for outperforming 3.1 Pro on coding and agentic tasks while being four times faster and often less than half the cost, with a Pro version coming soon.

Sam Altman: Altman: OpenAI will expand compute offerings as fast as possible

Sam Altman said OpenAI will continue a program until capacity runs out while preserving resources for ChatGPT and Codex, with plans to relaunch later.

Demis Hassabis: Gemini Omni enables major leap in multimodal world understanding

Google debuts Gemini Omni, a multimodal AI system that creates entirely new scenes from photos, video, and audio inputs, starting with video generation.

Demis Hassabis: Gemini Omni debuts with multimodal scene generation from any media

Hassabis unveils Gemini Omni, a multimodal model that edits and generates new scenes from photos, video, and audio inputs.

Demis Hassabis: Gemini Omni launches with advanced multimodal video editing

Hassabis introduces Gemini Omni, a multimodal AI system that generates new scenes from photos, video, and audio inputs, starting with video generation and editing features.

Andrej Karpathy: Andrej Karpathy joins Anthropic to return to LLM R&D

Karpathy announces he has joined Anthropic, expressing excitement to work on LLM research while maintaining his passion for education.

Andrej Karpathy: Andrej Karpathy joins Anthropic to return to R&D

Andrej Karpathy announces he has joined Anthropic, expressing excitement about returning to R&D at the frontier of LLMs while maintaining his passion for education.

Andrej Karpathy: Andrej Karpathy joins Anthropic as MTS to focus on LLM R&D

Karpathy announces joining Anthropic as an MTS to work on LLM research while maintaining his passion for education.

Andrej Karpathy: Andrej Karpathy joins Anthropic to return to LLM R&D

Karpathy announces he has joined Anthropic, expressing excitement about returning to LLM research and development.

Demis Hassabis: Demis Hassabis declares he is 'locked in'

The DeepMind CEO posts a cryptic but enthusiastic one-liner signaling intense focus just ahead of a wave of major product announcements.

Andrej Karpathy: 8xH100 cloud GPU requirement would stall any tutorial

Karpathy jokes that tutorial videos requiring an 8xH100 setup would immediately stall viewers due to the hardware accessibility barrier.

Andrej Karpathy: Realized 8XH100 requirement would block viewers on step 1

Karpathy notes that requiring an 8-GPU H100 setup would immediately prevent most viewers from following his nanochat tutorial.

Andrej Karpathy: Cloud GPU access is the new step-one blocker for AI tutorials

Karpathy observes that requiring eight H100 GPUs for a tutorial would immediately prevent most viewers from completing step one.

Demis Hassabis: Google DeepMind demos AI-powered intelligent mouse pointer

Demis Hassabis shares a prototype from his team that reimagines the mouse pointer with integrated AI intelligence, available to try in Google AI Studio.

Demis Hassabis: DeepMind unveils prototype for intelligent AI mouse pointer

Google DeepMind demonstrates an AI-powered intelligent mouse pointer prototype that users can already try out inside Google AI Studio.

Demis Hassabis: Isomorphic Labs raises $2.1B to turbocharge AI drug discovery

Demis Hassabis announces $2.1 billion in new funding for Isomorphic Labs to advance its mission of using AI to reimagine drug discovery and tackle disease.

Demis Hassabis: Isomorphic Labs raises $2.1B to turbocharge AI drug discovery

Isomorphic Labs secures $2.1 billion in new funding to accelerate its mission of using artificial intelligence to reimagine drug discovery and combat disease.

Andrej Karpathy: Ask LLMs for HTML output—it's the next interface frontier

Karpathy suggests requesting HTML from LLMs for richer outputs, predicting a progression from text to markdown to HTML and eventually interactive neural videos.

Andrej Karpathy: Karpathy predicts AI output will evolve from text to neural video

Karpathy argues that AI outputs should progress from markdown to HTML and eventually to interactive neural videos, while audio remains the ideal human input.

Andrej Karpathy: Ask LLMs for HTML output and the future of AI interfaces

Karpathy suggests requesting HTML from LLMs as a better interface format and predicts AI outputs will evolve toward rich, interactive visual experiences.

Andrej Karpathy: Karpathy: The future of LLM output is HTML, then neural video

Andrej Karpathy suggests prompting LLMs to return HTML for richer output, predicting a progression from text and markdown toward interactive neural video as the ideal AI-to-human interface.

Andrej Karpathy: Future of AI output: from markdown to HTML to neural video

Karpathy advises prompting LLMs to return HTML for richer output, arguing that vision is the preferred AI output format and interfaces will evolve toward interactive media.

Andrej Karpathy: Karpathy: HTML output is the next frontier for LLM interfaces

Andrej Karpathy argues that AI output should evolve from text to HTML and eventually neural video, while audio remains the ideal input modality.

Andrej Karpathy: AI I/O should evolve from markdown to HTML to neural video

Karpathy argues that LLM outputs should progress from markdown to HTML and eventually to interactive neural videos, while audio remains the ideal human input.