Inside the suddenly explosive world of AI safety | The Verge

The Verge has published an article titled 'Inside the suddenly explosive world of AI safety,' examining the state of the AI safety research field. The piece focuses in part on METR, the organization whose third-party research into AI risk 'now inspires fear in leading AI labs,' according to the excerpt. According to the article, METR started with just two people. The excerpt describes the organization as having grown from that two-person origin to a position of influence with leading AI labs.

The article also recounts the departures of safety leaders at OpenAI. The excerpt names Ilya Sutskever and Jan Leike among them. It further reports that Anthropic's head of safeguards research departed in February. According to the excerpt, that individual penned an open letter alleging that 'the world is in peril.' The excerpt does not state a reason for the OpenAI departures.

The Verge article groups these developments into a single account of the AI safety research field, linking a third-party evaluator's growing influence to the departure of senior safety staff from two leading labs. The excerpt applies the phrase 'suddenly explosive' to the field. It attributes the fear METR inspires in leading AI labs to its third-party research into AI risk, rather than to work conducted inside the labs themselves. The departures it cites span OpenAI and Anthropic. The open letter it cites comes from the head of safeguards research at Anthropic, who left in February.