Employees at leading artificial intelligence companies and research institutions are experiencing significant mental health challenges connected to concerns over the potential societal impact of their work. Staff at Britain’s AI Security Institute (AIsi), known for its role in testing cutting-edge AI models, have reportedly been placed on stress leave and are receiving counselling, according to sources familiar with the situation. Similar issues have been reported by employees at major AI firms including Anthropic, OpenAI, and Google DeepMind.

Insiders attribute the rising stress levels to demanding schedules for testing AI systems and widespread unease about the rapid advancement and deployment of these technologies. Several senior researchers have resigned, citing the emotional toll and apprehensions about releasing AI models they consider unsafe. A DeepMind employee described the effort to enhance AI safety as “an impossible task,” noting a pervasive sense of being overwhelmed and fear that the technology might cause serious harm.

Concerns extend beyond individual companies to broader industry dynamics, with some experts warning of reckless AI development driven by competitive pressures among corporations. Researchers have highlighted the increasing use of AI in sensitive areas such as energy infrastructure and conflict zones, raising alarms about insufficient regulatory and corporate safeguards. According to multiple industry voices, this combination exacerbates anxiety among those working with the technology.

Former and current AI researchers have pointed to particularly acute worries in specialized fields like cybersecurity and biochemistry, where AI’s growing capabilities include the potential creation of novel viruses. One former AIsi researcher expressed ongoing anxiety about whether effective controls could be put in place or if humanity is witnessing an uncontrolled escalation of risk.

This month, Jacob Coxon, who previously worked at Anthropic and OpenAI, resigned amid concerns that these organizations are “gambling” with the future of humankind. Additionally, Chi Onwurah, chair of the UK House of Commons science and technology committee, described the culture within some tech companies as “toxic for developers of good conscience,” emphasizing the moral conflict experienced by many AI practitioners.

The developments raise urgent questions about how AI companies address employee wellbeing and ethical accountability while advancing technologies that may profoundly impact society at large. As tensions around the pace and direction of AI progress intensify, the mental health of those at the forefront of this field underscores the complex challenges facing the AI industry and its governance.