Descriptions:
Nate Herk unpacks one of the stranger moments in recent AI history: OpenAI and Anthropic — the two companies racing hardest to build more powerful AI — have both simultaneously published documents calling for international coordination to slow down AI development, while also filing to go public. OpenAI’s “Built to Benefit Everyone” plan, co-authored by Sam Altman and their chief scientists, sets out three goals: an AI-automated research system by March 2028, broadly shared economic gains, and an international body empowered to take “coordinated action, including slowing frontier development when needed.” Anthropic’s report, published around the same time, calls for a verifiable pause mechanism if AI becomes too dangerous.
Herk explores the obvious tension: why would the companies with the most to gain from speed voluntarily ask for brakes? His answer is that they can’t slow down unilaterally — the competitive dynamics of the AI race make any single actor who pauses a loser — so they’re trying to create external enforcement that applies equally to everyone. He draws an analogy to nuclear arms control, noting that training frontier models leaves a visible footprint: massive, traceable chip orders from a single global supply chain, enormous power draws, and large physical construction that is difficult to hide.
The video is aimed at general audiences and takes a deliberately non-partisan tone, framing AI as an amplifier without inherent moral valence. Herk argues that what the labs are really asking for is societal preparation time — not a pessimistic view of the technology — and that the gap between people building these systems and the public who will live with them is the central challenge of the moment.
📺 Source: Nate Herk | AI Automation · Published June 16, 2026
🏷️ Format: Opinion Editorial
