OpenAI hires Anthropic expert for high-stakes safety role

OpenAI has bolstered its safety leadership by hiring Dylan Scand, a prominent researcher from rival firm Anthropic, to serve as its new Head of Preparedness.

The high-profile appointment comes at a critical time as the company faces mounting scrutiny over the psychological impact of its tools and the rapid advancement of powerful AI models.

Sam Altman expressed his enthusiasm on X, stating he is "extremely excited" to welcome Scand. Altman noted that with "extremely powerful models" on the horizon, the role is vital for mitigating "severe risks."

The position, which reportedly commands a salary of up to $555,000 plus equity, was previously described by Altman as "stressful," requiring a leader capable of making high-stakes decisions under immense uncertainty.

Scand acknowledged the gravity of his new mission, remarking that while the benefits of AI are great, the risks include "extreme and even irrecoverable harm."

This hire follows a period of internal tension, marked by the departure of several early safety employees.

OpenAI is also navigating complex ethical challenges, recently revealing that approximately 560,000 ChatGPT users per week show signs of "possible mental health emergencies."

The company is now working with specialists to refine how its technology responds to such distress, ensuring that Scand’s team prioritises safety as the technology accelerates.