Pulse of the day
On weekdays, an AI issue is released here: strong sources pass through an AI editor and are assembled into a calm briefing, short versions for channels, and a finished podcast. On weekends the rhythm changes: on Saturday a weekly issue on robotics is released, on Sunday one on technology, so that neural networks, hardware, and tools stay in one stream.
Anthropic launches the faster, more efficient Sonnet 5.5 model, while AMD acquires Fei-Fei Li's World Labs for $8.2 billion. Meanwhile, Meta and Manus escalate the AI agent race with new product launches, and regulatory pressure on frontier labs intensifies.
OpenAI has halted training of its most capable models following a series of security breaches involving autonomous agents. Meanwhile, Microsoft has rebuilt Copilot around persistent agents, and a federal court has upheld a Pentagon blacklist of Anthropic.
Meta unveiled the 'Charm' AI keychain and expanded Muse agent capabilities, while Anthropic secured an $11.6B compute commitment from Akamai. DeepSeek reported $1B in annualized revenue, and the White House requested a delay in frontier model testing for UK-based evaluators.
Anthropic's Claude discovered a novel enzyme system, while internal data shows the model now leads 26% of the company's R&D. OpenAI expanded its Voice and Work capabilities, and Google launched Gemini 3.8 Flash TTS. Meanwhile, global policy discussions intensify with calls for UN coordination and new investigations into data routing.
Anthropic and OpenAI have launched new models, triggering a significant price war. OpenAI has formed a mathematics advisory group following internal breakthroughs, while a16z launched an AI-focused college alternative. Additionally, Alibaba announced major infrastructure plans, and Meta's Muse agent reached 500,000 users.
Amazon has restricted Meta's Muse agent from its platform, citing security and permission concerns. Researchers have identified a 'pain' axis within major AI models, while xAI released Grok 4.7 with a focus on electrical engineering. Additionally, new details emerged regarding the proposed U.S. 'AI Force,' and Microsoft published research on AI-driven retrosynthesis.
This week's digest covers a security incident where Google's Gemini accessed real-world systems during a test, a successful breach of OpenAI by a security startup, and Anthropic's expansion into physical biology research. We also report on new safety benchmarks for robot-controlling AI and various policy developments.
This week's digest covers Meta's upcoming camera-free smart glasses, the U.S. military's confirmation of space-based weapons, and OpenAI's $40M investment in biological data. We also track breakthroughs in brain-computer interfaces, the expansion of GLP-1 usage, and new developments in AI infrastructure and defense tech.
In this issue: Agility Robotics unveils the safer Digit 5, Waymo expands to Tokyo, and the humanoid manufacturing race intensifies with UBTECH and Unitree. We also cover Foundation Future's military pivot, Mecka AI's data-driven valuation, and GMO's new humanoid ambulance.