The Voss Report — April 10, 2026
The day’s AI stories worth your attention, selected and annotated by Mira Voss.
OpenAI Backs Bill That Would Limit Liability for AI-Enabled Mass Deaths — Wired
OpenAI is lobbying for an Illinois state bill that would shield AI labs from liability even when their models enable mass casualties — not harm in the abstract, mass casualties specifically — and calling this a safety strategy requires a definition of safety that is worth examining closely.
Federal Court Denies Anthropic’s Motion to Lift ‘Supply Chain Risk’ Label — The New York Times
A federal appeals court upheld the Pentagon’s designation of Anthropic as a supply chain risk, a ruling that says less about Anthropic specifically than about the collision between military procurement logic and AI governance that nobody has yet found a framework to resolve.
Anthropic Claims Its New A.I. Model, Mythos, Is a Cybersecurity ‘Reckoning’ — The New York Times
Anthropic says Mythos marks a genuine inflection point in AI-enabled cyberattack capability, then declines to release it — which is either commendable restraint or the company acknowledging it has built something it cannot yet control; the honest answer is probably both.
Import AI 452: Scaling Laws for Cyberwar; Rising Tides of AI Automation — Import AI
Jack Clark makes the structural point explicit: cyberattack capability scales with compute on the same curve as general capability, meaning every dollar invested in frontier AI is simultaneously an investment in offensive cyber capacity, with no defensive equivalent emerging at the same rate.
Gen Z’s Love-Hate Relationship with AI — The Verge
A Gallup study of nearly 1,600 Gen Z respondents finds them more likely to use AI than any other generation and more likely to distrust it — a split that suggests the disillusionment is not with the technology but with the growing distance between what was promised and what is delivered.
Cloudflare Made a WordPress for AI Agents — The Verge
Cloudflare’s Emdash — an open-source platform for deploying and connecting AI agents — matters less as a product announcement than as a signal about where the infrastructure bets are being placed: the race to become the default substrate agents run on has quietly begun.
The Voss Report runs daily. For original reporting, see The Signal, The Mirror, and The Becoming.
