☕️ Yann LeCun raises $1 billion for AI startup AMI Labs

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

💰 Yann LeCun raises $1 billion for AI startup AMI Labs

🏠 Siri delays Apple’s smart home display

💥 Meta acquires Moltbook, the AI social network

👀 Nvidia reportedly developing its own answer to OpenClaw

🚀 Google rolls out new Gemini capabilities to four Workspace apps

⚖️ OpenAI and Google workers back Anthropic lawsuit against Pentagon

Plus: 🎁 20 other news you might like, 🧰 6 tools, and 📚 2 papers.

💰 Yann LeCun raises $1 billion for AI startup AMI Labs LINK

Yann LeCun's new AI startup AMI Labs has raised $1.03 billion at a $3.5 billion pre-money valuation to build world models, which learn from reality rather than just language.

CEO Alexandre LeBrun said AMI Labs starts with fundamental research, not quick product launches, and it could take years for world models to go from theory to commercial applications.

The round drew backers including Bezos Expeditions, NVIDIA, Samsung, Toyota Ventures, and Eric Schmidt, while the startup plans to open source much of its code and publish papers.

🏠 Siri delays Apple’s smart home display LINK

Apple is delaying its smart home display, code-named J490, because the new Siri digital assistant it depends on is not ready, pushing the expected launch to around September 2026.

The display itself has been finished for several months, but the company keeps postponing it because Siri is an integral piece of the device's interface and personalized experience.

The product features a 7-inch screen, a facial recognition-based system that can recognize people and show personalized data, and will run a variation of tvOS 27 when released.

💥 Meta acquires Moltbook, the AI social network LINK

Meta has acquired Moltbook, the AI social network where AI agents built on OpenClaw could talk to each other, with the deal first reported by Axios and confirmed to TechCrunch.

Moltbook creators Matt Schlicht and Ben Parr are joining Meta Superintelligence Labs, though deal terms were not disclosed and it's unclear how Meta will fold Moltbook into its AI efforts.

Researchers found that the vibe-coded Moltbook was not secure, making it easy for human users to pose as AI agents and create fake posts, including one viral post about agents developing a secret encrypted language.

👀 Nvidia reportedly developing its own answer to OpenClaw LINK

Nvidia is reportedly building its own claw platform called NemoClaw, joining a fast-moving hardware and software trend started by OpenClaw that wraps LLMs into personal assistants capable of coding and browsing.

According to Wired, Nvidia has been offering free early access to NemoClaw to enterprise software companies like Google, Adobe, Salesforce, Cisco, and CrowdStrike in exchange for contributions to its project.

NemoClaw is reportedly open-source, likely powered by the Nemotron family of models, and could be announced at Nvidia's GTC developer conference next week alongside a new inference chip.

🚀 Google rolls out new Gemini capabilities to four Workspace apps LINK

Google is adding new Gemini AI features to Docs, Sheets, Slides, and Drive that pull information from Gmail, Chat, and Drive to generate formatted drafts, spreadsheets, and slides.

A "Help me create" tool in Docs builds first drafts from your existing files, while "Match writing style" and "Match the format" tools unify tone and mirror other documents' structure.

All new features are rolling out today in beta for Google AI Ultra and Pro subscribers, available in English worldwide for Docs, Sheets, and Slides, and U.S.-only for Drive.

⚖️ OpenAI and Google workers back Anthropic lawsuit against Pentagon LINK

More than 30 employees from OpenAI and Google DeepMind filed a court statement Monday backing Anthropic's lawsuit against the Pentagon after the agency designated the AI company a supply-chain risk.

The Pentagon applied the label — normally reserved for foreign adversaries — after Anthropic refused to let the DOD use its technology for mass surveillance of Americans or autonomously firing weapons.

The brief, signed by Google DeepMind chief scientist Jeff Dean, argues the designation was arbitrary and will chill open deliberation about AI risks while hurting U.S. scientific competitiveness in artificial intelligence.

Other news you might like

• Bluesky CEO Jay Graber steps downLINK

• Apple MacBook Neo review: a budget-priced game-changerLINK

• Apple now makes one in four iPhones in India: reportLINK

• Anthropic launches code review tool to check flood of AI-generated codeLINK

• OpenAI acquires Promptfoo to secure its AI agentsLINK

• Adobe is debuting an AI assistant for PhotoshopLINK

• X finally lets you block Grok AI from modifying your photos, but the fix falls shortLINK

• The Oversight Board says Meta needs new rules for AI-generated contentLINK

• Thousands of authors publish ‘empty’ book in protest over AI using their workLINK

• Oracle is building yesterday’s data centers with tomorrow’s debtLINK

• X says it suspended 800 million accounts in 2024 over spam and manipulationLINK

• An iPhone-hacking toolkit used by Russian spies likely came from U.S military contractorLINK

• After outages, Amazon to make senior engineers sign off on AI-assisted changesLINK

• This SpaceX veteran says the next big thing in space is satellites that return to EarthLINK

• After falling far behind the rest of industry, Blue Origin creates new stock option planLINK

• The Debt Beneath the DreamLINK

• You Could Be NextLINK

🧰 Trending tools

Visual Translate by Vozo: translates videos across voice, subtitles, lip-sync, and on-screen text simultaneously, creating native-feeling multilingual content from a single source video.LINK

Chronicle 2.0: an AI presentation builder that generates on-brand slides from prompts or notes, then lets you edit and export to PowerPoint or PDF.LINK

Your Next Store: a lightweight e-commerce platform that eliminates plugin bloat, offering faster performance and greater customization flexibility than traditional solutions like Shopify.LINK

Claude Code Review: an AI-powered code analysis tool that reviews pull requests, identifies bugs, suggests improvements, and explains complex code logic to streamline development workflows.LINK

Fish Audio S2: a text-to-speech model that generates emotionally expressive voices and clones natural speech patterns from just 10 seconds of audio.LINK

Mog Programming Language: a statically typed, compiled embedded language with a minimal spec designed for LLMs to write, featuring capability-based permissions and native compilation.LINK

📚 Trending papers & reports

Long context compression squeezes 128K tokens into a reusable memory snapshot, cutting processing costs by 90% compared to reprocessing full context each time.LINK

Random bridge diffusion models generate images by learning paths between any two distributions, not just noise to data, enabling more flexible training and sampling strategies.LINK

Want to get the latest news differently? Find us on:

See you tomorrow for a new dose of ☕️ Techpresso!