AI researchers at top labs take stress leave over fears of dangerous technology

A recent report shows staff at major AI developers—OpenAI, Anthropic, Google DeepMind—and the UK-funded AI Security Institute are increasingly taking stress-related medical leave and receiving counselling. Insiders describe low morale driven by relentless model releases, tight testing schedules, and the rapid expansion of AI capabilities, particularly in cyberwarfare and biochemistry, which they fear could be exploited by rogue actors.

At AISI, multiple employees have been signed off work, while DeepMind workers cite mental strain over uncertain future uses, especially after a Pentagon contract. High-profile resignations, such as Jacob Coxon's departure from Anthropic and Mrinank Sharma's exit after a public warning, highlight deep unease about the sector's profit-driven pace. Leaders including Dario Amodei, Sam Altman, Demis Hassabis and Elon Musk have called for a slowdown, yet staff attrition and anxiety remain significant concerns.