Mario Gabriele interviews Skype co-founder Jaan Tallinn, a longtime AI risk advocate and early Anthropic investor, on his evolving risk estimate, his call for a frontier training moratorium, and hardware-based verification for AI governance. The post is a podcast show-notes and episode summary.
▶
NYU math professor Tristan Buckmaster says a Tuesday release of 722 AI-generated math papers from OpenAI wiped out entire research programs and damaged early-career mathematicians' careers. The claim is made in a video clip.
Anjney Midha announces hundreds of MI355x and B300 nodes are live on nationalcompute.com, subsidized for .edu and .gov users. The site describes a public-sector compute access program with a capacity tracker.
▶
Trung Phan recalls a DeepMind documentary showing Demis Hassabis's reaction when told AlphaFold could predict all known protein sequences, then its release. He quotes a Bloomberg report that Isomorphic Labs is in early talks to raise funds at a valuation of at least $40 billion.
Bloomberg @businessIsomorphic Labs, an AI-powered drug discovery startup spun out of Alphabet’s Google DeepMind, is in early talks to raise new funds at a valuation of at least $40 billion, according to people familiar with the effort. bloomberg.com/news/articles/2026-10-08/alphabe…
▶
Claude announced that Claude Dashboards and Claude Motion are now in beta. Users can ask Claude to turn data into live dashboards and ideas into animated explainers.
OpenAI announced it is releasing a range of new mathematical results produced by an internal frontier model, consulting the Institute for Advanced Study's Advisory Group on Mathematics and Artificial Intelligence on how to release them, with materials published on GitHub.
Tony Dinh comments that the new service supports image input, quoting an OpenAI Developers announcement of the Decisions API. The API lets apps choose models, tools or actions in near real time and is claimed to be up to 10x faster than GPT-6 Luna via the Responses API.
OpenAI Developers @OpenAIDevsLet your app choose the right model, tool, or action in near real-time with Decisions API, now available to all developers in public beta.
The Decisions API makes decisions up to 10x faster than GPT-6 Luna through the Responses API.
