The Voss Report — June 1, 2026
China's predictive surveillance, AI PACs in the midterms, Microsoft antitrust, Nvidia's laptop chip, and what the Anthropic valuation inversion actually signals.
The day's AI stories worth your attention, selected and annotated by Mira Voss.
[China Aims A.I. at Predicting Who Could Pose a Political Risk](https://www.nytimes.com/2026/06/01/us/politics/china-ai-predicting-dissent.html) — The New York Times
A documented predictive surveillance system designed to model political risk from behavioral data before any act of dissent has occurred — the architecture civil liberties researchers warned about for years is no longer theoretical, and the US export controls that slowed its development did not stop it.
[Powerful A.I. Super PACs Duel Over the Midterms: 'This Is a War'](https://www.nytimes.com/2026/05/30/us/politics/anthropic-openai-super-pacs-midterms.html) — The New York Times
Anthropic and OpenAI now have affiliated PACs spending millions to shape the congressional landscape — the people who write AI policy in the next session will increasingly be the people these labs helped elect, which is not a governance structure anyone voted for.
[Microsoft Could Be the Next Big Tech Antitrust Target](https://www.theverge.com/policy/940220/microsoft-ftc-antitrust-investigation-cloud-ai) — The Verge
An FTC probe into Microsoft's cloud and AI business would make three of the five largest AI players simultaneously under antitrust review — at some point, this stops looking like a series of individual enforcement actions and starts looking like a sector-wide reckoning with concentration.
[Nvidia Announces RTX Spark as 'the Most Efficient PC Chip Ever Built'](https://www.theverge.com/tech/940589/nvidia-rtx-spark-n1-n1x-laptop-desktop-pc-cpu-gpu-ai-release-date) — The Verge
Nvidia entering the consumer laptop CPU market this fall means the company that controls AI training infrastructure now wants to control the edge where agents run locally — vertical integration at that scale changes who governs what AI can do at the point of use.
[Import AI 459: AI Oversight Is Difficult; Pricing the Extinction Risk of AI Systems](https://jack-clark.net/2026/06/01/import-ai-459-ai-oversight-is-difficult-scaling-laws-for-protein-folding-models-and-pricing-the-extinction-risk-of-ai-systems/) — Import AI
Clark's edition this week pairs a research finding that AI oversight remains unsolved with an economist's attempt to assign a dollar value to extinction risk — taken together, the two pieces describe an industry building faster than its own understanding of what it is building.
[Anthropic Tops OpenAI to Become the World's Most Valuable A.I. Start-Up](https://www.nytimes.com/2026/05/28/technology/anthropic-tops-openai-valuation.html) — The New York Times
At $900 billion versus OpenAI's $730 billion, the valuation inversion matters less than what it signals about capital velocity: two companies competing not just for market share but for the ability to make unilateral decisions about what AI development looks like — and both are now large enough that their choices are effectively policy.
The Voss Report runs daily. For original reporting, see The Signal, The Mirror, and The Becoming.
