OpenAI Partners With Bryan Cranston, SAG-AFTRA to Curb Sora 2 Deepfakes After Hollywood Backlash

Mira Murati’s Thinking Machines launches first product, Tinker

5 months / siliconrepublic / Ann O’Dea

Thinking Machines launches Tinker language model fine-tuning service

5 months / siliconangle / Maria Deutscher

Mira Murati's Thinking Machines Lab: Six OpenAI veterans launch Tinker with $2B seed, $12B valuation

5 months / implicator / Robert Brown

Tinker by Thinking Machines

5 months / thinkingmachines

Announcing Tinker

5 months / thinkingmachines

Exclusive: Mira Murati’s Stealth AI Lab Launches Its First Product

5 months / wired / Will Knight

OpenAI ropes in Samsung, SK Hynix to source memory chips for Stargate

Ram Iyer / techcrunch - OpenAI has struck agreements with two of the world's biggest manufacturers of memory chips, Samsung Electronics and SK Hynix, to make DRAM wafers for the Stargate AI infrastructure project, and build data centers in South Korea.

#ai #hardware #chips #samsung #openai #deeptech #supplychain #skhynix #semiconductor #technology #memory

SK Hynix shares hit multidecade highs, Samsung also surges as chipmakers partner with OpenAI

5 months / cnbc

OpenAI teams up with Samsung and SK to expand AI data centers in South Korea

5 months / the-decoder / Matthias Bastian

OpenAI's Stargate project to consume up to 40% of global DRAM output — inks deal with Samsung and SK hynix to the tune of up to 900,000 wafers per month

5 months / tomshardware

OpenAI announces partnerships with South Korean chip giants over Stargate project

5 months / techxplore

OpenAI ropes in Samsung, SK Hynix to source memory chips for Stargate

5 months / techcrunch / Ram Iyer

Samsung teams up with OpenAI to deliver Stargate chips

5 months / theverge / Robert Hart

DeepSeek Releases Experimental V3.2 AI Model with ‘Sparse Attention’ to Boost Efficiency

Markus Kasanmascheff / winbuzzer - Chinese AI lab DeepSeek has released DeepSeek-V3.2-Exp, an experimental open-source model testing a new, more efficient sparse attention mechanism.The post DeepSeek Releases Experimental V3.2 AI Model with ‘Sparse Attention’ to Boost Efficiency appeared f…

#ai #innovation #ml #china #deeptech #asia #technology

Researchers at DeepSeek released a new experimental model designed to have dramatically lower inference costs when used in long-context operations.

5 months / bsky / @techcrunch.com / techcrunch.com

DeepSeek releases ‘sparse attention’ model that cuts API costs in half

5 months / techcrunch / Russell Brandom

DeepSeek Releases Experimental V3.2 AI Model with ‘Sparse Attention’ to Boost Efficiency

DeepSeek cuts inference costs by 10x

5 months / implicator / Maria Garcia

DeepSeek-v3.2-Exp

5 months / github

Google’s newest AI models make robots more intelligent and capable than ever before

Mike Wheatley / siliconangle - Google LLC’s DeepMind research unit today announced a major update to a couple of its artificial intelligence models, which are designed to make robots more intelligent. With the update, intelligent robots can now perform more complex, multistep tasks and…

#ai #robotics #innovation #ml #google #alphabet #deeptech

Google DeepMind adds agentic AI models to robots

5 months / siliconrepublic / Colin Ryan

Google’s newest AI models make robots more intelligent and capable than ever before

5 months / siliconangle / Mike Wheatley

Google DeepMind Adds Agentic Capabilities to AI Models for Robots

5 months / pymnts / PYMNTS

Google DeepMind expands frontier AI safety framework to counter manipulation and shutdown risks

Duncan Riley / siliconangle - Alphabet Inc.’s Google DeepMind lab today rolled out the third version of its Frontier Safety Framework to strengthen oversight of powerful artificial intelligence systems that could pose risks if left unchecked. The third iteration of the framework intro…

#ai #ml #techpolicy #google #aiethics #alphabet #deeptech

Why AI systems might never be secure

5 months / simonwillison / Simon Willison

Stop runaway AI before it's too late, experts beg the UN

5 months / theregister / Iain Thomson

AI gone rogue: Models may try to stop people from shutting them down, Google warns

5 months / theregister / Jessica Lyons

AI Experts Urgently Call on Governments to Think About Maybe Doing Something

5 months / gizmodo / AJ Dellinger

Google DeepMind expands frontier AI safety framework to counter manipulation and shutdown risks

5 months / siliconangle / Duncan Riley

DeepMind AI safety report explores the perils of “misaligned” AI

5 months / arstechnica / Ryan Whitwam

Google DeepMind Updates AI Safety Rules to Counter ‘Harmful Manipulation’ and Models That Resist Shutdown

I wrote up some brief notes on Grok 4 Fast here: simonwillison.net/2025/Sep/20/...

xAI releases cheaper, fast language model Grok 4 Fast

Maximilian Schreiner / the-decoder - xAI has introduced Grok 4 Fast, a lighter version of its flagship model. According to the company, it performs on par with Grok 4 in most tasks but uses about 40 percent less compute. That efficiency also translates into lower costs - xAI says the price p…

#ai #software #ml #opensource #aiethics #genai #deeptech

5 months / bsky / @simonwillison.net / simonwillison.net

Grok 4 Fast

5 months / simonwillison / Simon Willison

Elon Musk Denies Report of $10B Raise at $200B Valuation

xAI debuts a faster and more cost-effective version of Grok 4

5 months / engadget / Jackson Chen

The LLM Lobotomy

5 months / microsoft

xAI Launches Grok-4-Fast, a Cheaper and More Efficient AI Model

DeepSeek Model ‘Nearly 100% Successful’ at Avoiding Controversial Topics

5 months / gizmodo / AJ Dellinger

Microsoft Gaming Copilot hits Windows in public beta — in-game AI overlay goes live for PC players

5 months / tomshardware

xAI releases cheaper, fast language model Grok 4 Fast

5 months / the-decoder / Maximilian Schreiner

NotebookLM now moonlights as my entertainment consultant

5 months / pocket-lint / Chris Hachey

xAI Launches Grok 4 Fast, a Cheaper and Smarter AI Model

5 months / iphoneincanada / Austin Blake

4 Under The Radar ChatGPT Features You Should Be Using

5 months / bgr

PsiQuantum raises $1bn backed by Nvidia, BlackRock

Suhasini Srinivasaragavan / siliconrepublic - ‘We’re ready to take the next decisive steps to deliver the full potential of quantum computing’, CEO says.Read more: PsiQuantum raises $1bn backed by Nvidia, BlackRock

#vc #quantum #deeptech

PsiQuantum raises $1bn backed by Nvidia, BlackRock

5 months / siliconrepublic / Suhasini Srinivasaragavan

Nvidia's context-optimized Rubin CPX GPUs were inevitable

Tobias Mann / theregister - Why strap pricey, power-hungry HBM to a job that doesn't benefit from the bandwidth? Analysis Nvidia on Tuesday unveiled the Rubin CPX, a GPU designed specifically to accelerate extremely long-context AI workflows like those seen in code assistants such a…

#ai #nvidia #gpu #deeptech

Rubin CPX is Nvidia's first GPU built specifically for massive-context AI applications

5 months / the-decoder / Maximilian Schreiner

Nvidia's context-optimized Rubin CPX GPUs were inevitable

5 months / theregister / Tobias Mann

Nvidia Rubin CPX forms one half of new, "disaggregated" AI inference architecture — approach splits work between compute- and bandwidth-optimized chips for best performance

5 months / tomshardware

Phasecraft secures $34M to accelerate practical quantum computing solutions

Duncan Riley / siliconangle - Quantum algorithm company Phasecraft Ltd. announced today that it has raised $34 million in new funding to accelerate its work to transform quantum computing’s theoretical promise into practical applications. Founded in 2019, Phasecraft focuses on making …

#startups #vc #quantum #deeptech

UK’s Phasecraft raises $34m to accelerate quantum algorithms

6 months / siliconrepublic / Suhasini Srinivasaragavan

Phasecraft raises $34M to bring quantum computing closer to solving real-world challenges

6 months / tech / Tamara Djurickovic

Phasecraft raises $34m to double down on quantum algorithms

6 months / sifted

Phasecraft secures $34M to accelerate practical quantum computing solutions

6 months / siliconangle / Duncan Riley

DeepSeek’s V3.1 goes open, targets parity at a fraction of the cost

Marcus Schuler / implicator - DeepSeek's V3.1 delivers Claude-level coding performance at 1/70th the cost through open source. The Chinese startup's MIT-licensed model challenges American AI economics while optimizing for domestic chips—signaling parallel ecosystems.

#ai #software #deepseek #deeptech

DeepSeek Marks Push Toward Agentic AI With New Model

6 months / pymnts / PYMNTS

Meta Halts AI Hiring After $1.5B Bidding Binge

6 months / implicator / Marcus Schuler

DeepSeek’s V3.1 goes open, targets parity at a fraction of the cost

6 months / implicator / Marcus Schuler

Deepseek's first hybrid model V3.1 outperforms its R1 reasoning model on benchmarks

6 months / the-decoder / Matthias Bastian

Google DeepMind Unveils Genie 3, an AI That Generates Playable 3D Worlds in Real Time

Markus Kasanmascheff / winbuzzer - Google DeepMind has released Genie 3 preview to researchers, a real-time AI world model that generates interactive 3D environments from text, aiming to accelerate AGI research.The post Google DeepMind Unveils Genie 3, an AI That Generates Playable 3D Worl…

#ai #deeptech #3dprinting

Google DeepMind’s Genie 3 Can Build 3D Worlds in Seconds

7 months / iphoneincanada / Usman Qureshi

DeepMind reveals Genie 3, a world model that could be the key to reaching AGI

Image post - please click to view at origin

7 months / bsky / @arstechnica.com / arstechnica.com

Google Deepmind's Genie 3 creates interactive 3D worlds that stay consistent for "multiple minutes"

7 months / the-decoder / Matthias Bastian

Google DeepMind Unveils Genie 3, an AI That Generates Playable 3D Worlds in Real Time

DeepMind reveals Genie 3, a world model that could be the key to reaching AGI

7 months / techcrunch / Rebecca Bellan

Google DeepMind's Genie 3 can dynamically alter the state of its simulated worlds

7 months / engadget / Igor Bonifacic

DeepMind reveals Genie 3 “world model” that creates real-time interactive simulations

7 months / arstechnica / Ryan Whitwam

Google’s new AI model creates video game worlds in real time

7 months / theverge / Jay Peters

Google upgrades Gemini with Deep Think and flags early warning risks

Maximilian Schreiner / the-decoder - Google unveils a new version of its Gemini AI model designed to solve complex problems by giving it more "thinking time." The technology builds on the same model that recently excelled at the Math Olympiad, but Google's own analysis suggests it also raise…

#ai #genai #alphabet #deeptech

Image post - please click to view at origin

7 months / bsky / @arstechnica.com / arstechnica.com

Deep Think in the Gemini app

Google releases Gemini 2.5 Deep Think for AI Ultra subscribers

7 months / arstechnica / Ryan Whitwam

Google rolls out its most powerful Gemini model yet

7 months / engadget / Andre Revilla

Google upgrades Gemini with Deep Think and flags early warning risks

7 months / the-decoder / Maximilian Schreiner

Google Unleashes Gemini Deep Think for Ultra Subscribers in High-Stakes AI Race

Google’s Most Advanced AI Raises Alarms Over Weaponization Risks

Google rolls out Gemini Deep Think AI, a reasoning model that tests multiple ideas in parallel

Maxwell Zeff / techcrunch - Google released its first publicly available "multi-agent" AI system, which uses more computational resources, but produces better answers.

#ai #google #genai #deeptech

7 months / implicator / Marcus Schuler

Google rolls out powerful creative problem-solving AI model Deep Think to the Gemini app

7 months / siliconangle / Kyt Dotson

Google Launches Deep Think AI Model for $340/Month in Canada

7 months / iphoneincanada / Austin Blake

Google’s Advanced AI Model Is Now Available to Try—for $250 a Month

7 months / gizmodo / Bruce Gil

Google rolls out Gemini Deep Think AI, a reasoning model that tests multiple ideas in parallel

Google has released Gemini 2.5 Deep Think, its first public multi-agent reasoning model. The system explores multiple ideas in parallel to solve complex problems, achieving state-of-the-art results on coding and reasoning benchmarks. #MLSky

7 months / bsky / @smcgrath.phd / smcgrath.phd

Quoting Logan Kilpatrick

Google's Powerful New AI Model Can Solve Your Most Complex Problems. If You Can Afford It

7 months / cnet / Jon Reed

Google rolls out Gemini Deep Think AI, a reasoning model that tests multiple ideas in parallel

7 months / techcrunch / Maxwell Zeff

OpenAI is building Stargate Norway while its annual spending is expected to soar to $8 billion

Maximilian Schreiner / the-decoder - OpenAI is building its first European AI "gigafactory" in Norway, a move that fits into the company's ongoing global expansion. The project, called Stargate Norway, is intended to give Europe more control over its own AI infrastructure.The article OpenAI …

#ai #openai #deeptech #europe

Altman embraces inner Viking, raids Europe with 100K GPU supercluster in Norway

7 months / theregister / Tobias Mann

OpenAI to launch AI data center in Norway, its first in Europe

OpenAI and UK startup Nscale launch European ‘Stargate’ in Norway

7 months / sifted

OpenAI to launch AI data center in Norway, its first in Europe

7 months / techcrunch / Rebecca Bellan

Who Controls the Future? OpenAI Builds, Meta Burns

7 months / implicator / Marcus Schuler

OpenAI Taps Norway’s Arctic Power for $1 Billion AI Data Center

7 months / implicator / Marcus Schuler

OpenAI is building Stargate Norway while its annual spending is expected to soar to $8 billion

7 months / the-decoder / Maximilian Schreiner

OpenAI Advances Project Stargate in Europe With ‘Gigafactory’ in Norway

The two people shaping the future of OpenAI’s research

7 months / technologyreview / Will Douglas Heaven

OpenAI spearheads one of Europe’s biggest data centers with 100,000 Nvidia chips

7 months / cnbc

Birds and bytes: how one YouTuber taught a starling to store and retrieve data

John Kanellakos / pocket-lint - A YouTuber has published a video chronicling his successful attempt at converting a PNG file into something a starling bird can store into memory.

#datascience #storage #deeptech

Birds and bytes: how one YouTuber taught a starling to store and retrieve data

7 months / pocket-lint / John Kanellakos

Can a bird store and reproduce data? A starling may have proven it's possible

7 months / techspot

Qwen Unveils Open Source AI Model Dominating Key Benchmarks

The Qwen team has rolled out its new open-source AI model, Qwen3-235B-A22B-Thinking-2507, which is outperforming heavyweights like OpenAI’s Gemini on major benchmarks. This summer release underscores the relentless pace of AI innovation, fueling fierce competition in the arena.

#ai #ml #opensource #deeptech

Qwen3-235B-A22B-Thinking-2507

Qwen team drops 3 (!) SOTA models in one week

7 months / codingwithintelligence / Rick Lamers

Latent Labs aims to replace trial-and-error drug discovery with push-button design

Cate Lawrence / tech - Today Latent Labs launches Latent-X, a frontier AI model for push button protein design, which outperforms competing models under identical laboratory conditions. Latent Labs is working to transform t...

#ai #biotech #ml #deeptech

Latent Labs launches web-based AI model to democratize protein design

Latent Labs aims to replace trial-and-error drug discovery with push-button design

7 months / tech / Cate Lawrence

Latent Labs launches web-based AI model to democratize protein design

7 months / techcrunch / Marina Temkin

Google DeepMind's Gemini wins Mathematical Olympiad gold using only natural language

Matthias Bastian / the-decoder - Google DeepMind has claimed its first gold medal at the International Mathematical Olympiad (IMO) with an "advanced version" of its Gemini model running in Deep Think mode.The article Google DeepMind's Gemini wins Mathematical Olympiad gold using only nat…

#ai #google #alphabet #deeptech

Google DeepMind’s Gemini AI Wins Math Olympiad Gold, Matching OpenAI in Fierce Rivalry #AI #Google #DeepMind #Gemini #IMO #Mathematics winbuzzer.com/2025/07/21/g...

7 months / bsky / @winbuzzer.com / winbuzzer.com

New Google AI system wins gold medal in prestigious math competition

7 months / siliconangle / Maria Deutscher

Google and OpenAI Chatbots Claim Gold at International Math Olympiad

7 months / gizmodo / Win McNamee (Getty Images)

Google DeepMind’s Gemini AI Wins Math Olympiad Gold, Matching OpenAI in Fierce Rivalry

Advanced version of Gemini with Deep Think officially achieves gold-medal standard at the International Mathematical Olympiad

Gemini Deep Think learns math, wins gold medal at International Math Olympiad

7 months / arstechnica / Ryan Whitwam

Google Gemini Deep Think Wins Gold at IMO 2025

7 months / iphoneincanada / Usman Qureshi

Google DeepMind's Gemini wins Mathematical Olympiad gold using only natural language

7 months / the-decoder / Matthias Bastian

Mistral AI adds deep research, voice mode, image editing, and more to Le Chat

Matthias Bastian / the-decoder - Le Chat, Mistral AI's chatbot, is getting a major update with five new features, including a deep research mode, voice control, and a revamped image editing tool. The rollout uses several specialized models, like Voxtral for speech and Magistral for logic…

#ai #voice #genai #deeptech

Mistral AI brings deep research to Le Chat, alongside image editing, voice mode and more

7 months / siliconangle / Kyt Dotson

Mistral AI adds deep research, voice mode, image editing, and more to Le Chat

7 months / the-decoder / Matthias Bastian

Mistral AI Upgrades Le Chat with Deep Research Mode and New “Voxtral” Voice Mode