What we're reading, in the order we're reading it.
Anthony and Harry's working stream — links, charts, tweets, and short takes. The unprocessed inputs behind the briefings.
GenAI, agentic AI, and beyond: How autonomous systems are redefining enterprise DNA
In an autonomous enterprise, systems don’t just automate, they decide, act, and evolve. Learn what enables this transformation and what makes autonomy operationally viable, not just aspirational.
AI designs new superbug-killing antibiotics for gonorrhoea and MRSA
Two new potential drugs have been designed, by an artificial intelligence model, to kill gonorrhoea and MRSA
Apple trained an LLM to teach itself good interface design in SwiftUI
A group of Apple researchers took a very interesting approach to get an AI to teach itself how to build good interfaces in SwiftUI.
The AI Was Fed Sloppy Code. It Turned Into Something Evil.
The new science of “emergent misalignment” explores how PG-13 training data — insecure code, superstitious numbers or even extreme-sports advice — can open the door to AI’s dark side.
Meta report reveals “sensual conversations” AI chatbots can have with kids
The revelations have intensified bi-partisan scrutiny of the Mark Zuckerberg-owned, tech giant’s safety protocols.
Blackwell architecture gives Nvidia’s new small workstation GPUs the firepower to rival much larger systems at a fraction of the size
Nvidia RTX Pro 4000 SFF and Pro 2000 pack extreme AI capabilities into compact devices
Google unveils another huge AI spending spree – tech giant is splashing out $9 billion in Oklahoma
Google will invest billions in Oklahoma
Scientists Are Getting Seriously Worried That We’ve Already Hit Peak AI
Skepticism is mounting as the rate of progress for scalable artificial intelligence models appears to be slowing down.
The AI Industry Just Got Some Horrible News
Wall Street appears to be yanking its money out of one of the largest datacenter providers in the game, potentially an early warning sign.
Inside the Biden Administration’s Gamble to Freeze China’s AI Future
What really motivated the US government to ban Nvidia from selling powerful computer chips to China?
More than 200 geoglyphs discovered in Nazca Desert with help of AI
The newly found Nazca Lines include depictions of human sacrifice and a priest carrying a human head
Anthropic takes on OpenAI and Google with new Claude AI features designed for students and developers
Anthropic launches learning modes for Claude AI that guide users through step-by-step reasoning instead of providing direct answers, intensifying competition with OpenAI and Google in the booming AI education market.
AI startup Cohere valued at $6.8 billion in latest fundraising, hires Meta exec
Canadian AI startup Cohere was valued at $6.8 billion following its latest $500 million funding round, as it seeks to expand its market share in a highly competitive industry of...
DINOv3: Self-supervised learning for vision at unprecedented scale
DINOv3 scales the approach to create universal vision backbones that achieve absolute state-of-the-art performance across diverse domains.
When AI Starts Seeing What Doctors See
AI is moving from reading about patients to seeing them, integrating images, labs, and notes like a clinician.
Using generative AI, researchers design compounds that can kill drug-resistant bacteria
With help from artificial intelligence, MIT researchers designed novel antibiotics that can combat a drug-resistant form of Neisseria gonorrhoeae and multi-drug-resistant Staphylococcus aureus (MRSA).
Mind-reading AI can turn even imagined speech into spoken words
A brain-computer interface has enabled people with paralysis to turn their thoughts about speech into played out words, reducing the effort required to attempt speech with most similar technologies
😸 This AI could change a life
gm legends, happy Thursday. Here’s today’s lineup: Ally Solos Glasses put a voice assistant in lightweight frames so blind and low vision folks can rea
AWS launches AI agent marketplace with a hackathon and $100k in prizes for developers
SuperOps and AWS have a new initiative for businesses struggling to practically implement agents.
NSF and NVIDIA award Ai2 a combined $152M to support building a national level fully open AI ecosystem
Ai2 has been awarded a combined $152 million from the U.S. National Science Foundation (NSF) and NVIDIA as part of a jointly funded project to advance our research and develop...
AWorld: Dynamic Multi-Agent System with Stable Maneuvering for…
The rapid advancement of large language models (LLMs) has empowered intelligent agents to leverage diverse external tools for solving complex real-world problems. However, as agents increasingly…
Diffusion LLMs Can Do Faster-Than-AR Inference via Discrete…
Diffusion Large Language Models (dLLMs) have emerged as a promising alternative to autoregressive (AR) LLMs for text generation, with the potential to decode multiple tokens in a single iteration….
NSF and NVIDIA partnership enables Ai2 to develop fully open AI models to fuel U.S. scientific innovation
The U.S. National Science Foundation announced a partnership with NVIDIA to develop a set of artificial intelligence models that will transform the ability of America’s scientists to leverage AI…
Mechanistic understanding and validation of large AI models with SemanticLens
Nature Machine Intelligence – SemanticLens is a tool that embeds artificial intelligence model components (such as neurons) into a searchable, human-understandable space. This enables automated…
Convo Lang – The language of AI
Convo-Lang is an open source AI-native programming language and ecosystem designed specifically for building powerful, structured prompts and agent workflows for large language models (LLMs) like GPT-4, Claude, Llama, DeepSeek,...
ChatGPT 5 – When Your AI Friend Gets a Corporate Makeover
Why did ChatGPT 5 launch trigger user revolts while curated influencer videos praised it? The answer reveals a fundamental challenge the AI industry must solve.
Exclusive: US embeds trackers in AI chip shipments to catch diversions to China, sources say
U.S. authorities have secretly placed location tracking devices in targeted shipments of advanced chips they see as being at high risk of illegal diversion to China, according to two people...
Google pledges $9 billion to expand AI, cloud infrastructure in Oklahoma
Alphabet’s Google said on Wednesday it will spend an additional $9 billion in Oklahoma over the next two years to expand cloud and artificial intelligence infrastructure.
Research Update: Algorithmic vs. Holistic Evaluation
Many AI benchmarks use algorithmic scoring to evaluate how well AI systems perform on some set of tasks. However, AI systems often produce code that scores well but isn’t production-ready...
Google to spend $9 billion in Oklahoma to expand AI, cloud infrastructure
Alphabet’s Google said on Wednesday it will invest an additional $9 billion in Oklahoma over the next two years to expand cloud and artificial intelligence infrastructure.
Chrome As The Next AI Battleground? Experts Weigh In
Tech veterans and competitive strategists break down how a single browser could decide who controls search, spending and user intent in the age of AI.
What happens the day after superintelligence?
We could soon find ourselves deferring to AI assistants that botsplain our every experience in real time. Is this empowerment or deferral?
AI as Tool Creator: The Next Frontier in Knowledge Work
As I move into my second six months of using AI daily, I’m convinced its most overlooked role isn’t writing content—or even code—but creating tools. We all know how generative...
Nvidia Unveils Agentic AI, Physical Robotics Models
The company is also presenting a number of research papers on making physical AI training models more accurate and effective
AI2’s MolmoAct model ‘thinks in 3D’ to challenge Nvidia and Google in robotics AI
The Allen Institute of AI (Ai2)’s new physical AI model MolmoAct moves the needle for robots that can move freely in physical space.
AWS doubles up on enterprise AI with Anthropic, OpenAI deals
AWS has expanded its AI offerings with Anthropic’s Claude 4 on Bedrock and OpenAI’s open weight models on Bedrock and SageMaker.
Probing Power-Seeking in LLMs
TL;DR In this experiment, we study power-seeking behavior in LLMs. We fine-tune Qwen2.5-Coder-32B-Instruct on a narrow, 6,000 input–output pair datas…
US government to take 15% cut of AI chip sales to China
The U.S. government is expected to start getting 15% of the revenues that AI chip companies like Nvidia generate from AI chip sales to China.
OpenAI brings back GPT-4o after user revolt
After unpopular GPT-5 launch, OpenAI begins restoring optional access to previous AI models.