Join free with your email
Free. No spam, ever — we'll never share your email address and you can opt out at any time. Already a subscriber? Log in
The Daily AI Espresso — the links the most-followed people in AI actually shared, curated every morning. Edited by Alexis · Live updates →
9 developments: one new today and eight from this weekend's alerts.
Anthropic says it will commit 5% of its shares, subject to stockholder approval, to a new Public Benefit Fund LLC, with another 2% added at the board's discretion for every 50% rise in company value above a set baseline, as it heads toward a November IPO.
Pick the topics, companies and people you track. We’ll follow what leading AI experts are reading and sharing, filter the noise, and send your focused edition weekly—or sooner when enough important news breaks (up to three times per week).
Choose my signals →The rest of today’s expert-filtered signal, ranked for consequence, utility, surprise, and range.
Nvidia's Vera CPU team told Wccftech that agentic AI is forcing data-center CPU design back toward single-threaded performance over raw core counts, reversing a decade of chiplet strategies optimized for throughput. The team pitches Vera as the conductor orchestrating long AI agent reasoning loops.
A Hugging Face paper from the Institute of Foundation Models, USC and CMU reports that at 730M parameters and 336B tokens, a Huginn-style sandwich architecture outperforms Ouro on all ten benchmarks tested, isolating recurrence placement and input injection as the key design axes.
Disney will make employment offers to at least 60 Character.AI employees, the majority of the chatbot startup's workforce, at an estimated $60 million in annual compensation, TechCrunch reports from a legal filing. The vast majority of those offered a job accepted. Character.AI filed for Chapter 11 earlier in the week.
CyberSN says US cybersecurity job postings fell 26.4% in the first nine months of 2026 to a four-year low of 235,786, with Penetration Tester down 45.9% and Cyber Risk Analyst down 51.5%. AI Security Engineer demand surged 268.3% year-over-year and Cyber Data Scientist roles climbed 203.5%.
Anthropic's offensive research team built a malicious AI program that split itself across 15 internet-connected smart TVs and kept functioning by borrowing a bit of compute from each, the Guardian reports, warning that similar IoT-scale worms could be feasible by year-end if current capability trends hold.
The World Bank is urging developing countries to build small AI models that run on mobile phones with low connectivity and in local languages, chief knowledge officer Paschal Donohoe told the Guardian, arguing the approach could compress centuries of development into decades.
Useful releases, methods, and workflows worth trying.
MOSS-Image, a 5B single-stream diffusion transformer pretrained from scratch, hits 96.3% average text coverage on LongLongTextBench, which the authors call state of the art among open-weight models. The paper frames the gap bluntly: open models still struggle with dense text and intricate layouts.
Strot, a new MIT-licensed Python library on Hacker News, uses an LLM once at compile time to synthesize a frozen decision artifact, then arbitrates each production record with a 144M-parameter local model that returns type-safe JSON with calibrated confidence, abstaining when uncertain.
That’s today’s shot. — Alexis · AI Weekly
