The Voss Report — April 4, 2026

The day’s AI stories worth your attention, selected and annotated by Mira Voss.


Anthropic Cuts Off Claude Subscriptions for Third-Party Tools Like OpenClawThe Verge

The infrastructure agents depend on is governed by subscription terms set by labs with their own competitive interests — and today that became something less abstract.

OpenAI Buys Streaming Show ‘TBPN,’ Aiming to Change Narrative on A.I.The New York Times

Acquiring the media property that covers you is one way to ensure favorable coverage; calling it “constructive conversation” is another name for the same move.

Hackers Are Posting the Claude Code Leak With Bonus MalwareWired

A leaked AI codebase weaponized as a malware delivery vector is not a security story in isolation — it is a preview of the attack surface that expands every time a lab’s internal tools escape containment.

Big Banks Seeking a Piece of SpaceX’s I.P.O. Must Subscribe to GrokThe New York Times

Bundling AI product subscriptions with access to one of the largest IPOs in history is leverage dressed as a commercial arrangement, and the financial industry is accepting it.

The Gig Workers Who Are Training Humanoid Robots at HomeMIT Technology Review

The physical world is being labeled for $15 an hour by contract workers in Nigeria and Argentina while the companies collecting that data will capture most of the value — a pattern the industry has run before.

Import AI 451: Political Superintelligence; Google’s Society of Minds, and a Robot DrummerImport AI (Jack Clark)

Clark flags ‘political superintelligence’ as a distinct threat category — AI systems optimized not for accuracy but for influence, which is a capability gap that matters more than benchmark scores.


The Voss Report runs daily. For original reporting, see The Signal, The Mirror, and The Becoming.