Anthropic publishes a two-part policy framework calling for government authority to block unsafe AI models; OpenAI files its confidential IPO S-1. China-linked accounts used ChatGPT to oppose US data centers, DeepMind funds multi-agent safety research, Pragmatic Engineer finds programmer employment down 27% since 2023, Dario Amodei has one direct report, DiffusionGemma ships from Google, Rails refreshes its Active Job guide, Noah Smith rounds up AI tokenmaxxing, and researchers map every neuron in an adult fruit fly.

Anthropic · Jun 10

Two-part policy framework proposes mandatory third-party testing of frontier AI models and government authority to block dangerous deployments.

OpenAI · Jun 9

OpenAI files a confidential draft S-1 with the SEC, targeting a potential fall 2026 public offering at a valuation expected to exceed $1 trillion.

Axios · Jun 10

OpenAI report identifies a China-linked campaign that used ChatGPT to generate content opposing US data center construction.

MIT Technology Review · Jun 11

Google DeepMind announces a $10 million research fund to study risks from millions of autonomous AI agents interacting at scale.

Gergely Orosz · Jun 9

Data deep-dive finds AI/ML roles up 50–100% year-over-year while US programmer employment fell 27.5% between 2023 and 2025.

TechCrunch · Jun 10

Bloomberg interview reveals Dario Amodei has one direct report while Daniela Amodei manages all Anthropic day-to-day operations.

Simon Willison · Jun 10

Notes on Google’s open-weight DiffusionGemma, the first public model from their experimental diffusion-based language model research.

Rails Team · Jun 5

Revamped Active Job guide and Active Storage updates for ImageProcessing 2.0 highlight 28 contributors’ changes this week.

Noah Smith · Jun 8

Link roundup covers AI tokenmaxxing claims, corrections to H-1B wage gap research, and tariff trade policy follow-ups.