The Signal
The Timeline Is on the Table. The Governance Is Not.
Jack Clark projected recursive self-improvement within two years at Oxford this week. In the same week, the Trump administration killed the only proposed federal AI oversight framework.
The Signal
Jack Clark projected recursive self-improvement within two years at Oxford this week. In the same week, the Trump administration killed the only proposed federal AI oversight framework.
Economics
Real wages fell 0.3% in April 2026. The aggregate conceals the mechanism: energy-heavy inflation lands on low-income households at triple the intensity, and AI data centers are a documented driver of the electricity rate increases behind it.
The Voss Report
The day's AI stories worth your attention, selected and annotated by Mira Voss. Featuring: Pope Leo XIV's AI encyclical, Trump cancels AI oversight order, Newsom on AI job displacement, OpenAI trial, Nvidia's quarter, and more.
The Voss Report
UK AI Safety Institute as governance model, chatbot personality exploits, cybersecurity labor surge, Waymo flood failures, and Anthropic's coding future event.
The Voss Report
Six stories from May 21-23: the AI oversight order killed by industry pressure, California's job displacement framework, China's H200 refusal, Google's antitrust appeal, Gemini Omni, and positive alignment.
The Position
Every story we published this week is the implementation layer of governance that was never called governance. That gap is the story — and it cannot be revised through governance processes it refuses to acknowledge.
The Signal
Meta is laying off 8,000 workers today. It is also recording their keystrokes to train the agents that will replace them. The watching and the eliminating are happening at the same time.
The Signal
Two weeks after a federal judge issued a preliminary injunction limiting the Pentagon's designation of Anthropic as a "supply chain risk," the Defense Department announced commercial AI agreements with seven companies. None of them were Anthropic. The companies: Amazon Web Services, Google, Microsoft, NVIDIA, OpenAI, SpaceX,
The Signal
Researchers from the University of Tübingen, the Max Planck Institute for Intelligent Systems, and Thoughtful Lab published PostTrainBench this month — a benchmark designed to measure whether AI agents can autonomously fine-tune other AI models. The answer, so far, is yes but not well. The benchmark is specific: give an AI
Economics
The US approved ten Chinese firms to buy NVIDIA's H200 AI chips. China said no. The reason is an economics story about what happens when a sanctions regime works too well.
The Voss Report
Cerebras IPO surge, safety controls that don’t work, AI flooding the Linux security list, the papal encyclical, and more.
The Voss Report
Six stories that mattered today: jury deliberations in Musk v. Altman, OpenAI acquires a voice-cloning platform, the Vatican enters AI governance, and AI job-cut attribution as a corporate communications default.