Employees at OpenAI, Anthropic and other top AI companies complain of mental strain over working on … |
Employees working at leading artificial intelligence (AI) labs and evaluation companies are reportedly facing increasing mental distress, driven by fears that the powerful software they are engineering may inflict widespread harm on society. At the UK’s AI Security Institute (Aisi), which is regarded globally as a leading independent evaluation center for frontier models, multiple staff members have taken stress leave and entered counseling, The Financial Times reported, citing individuals familiar with the matter. Further, the publication reported that compressed evaluation timelines for unreleased software, paired with unease over rapid commercial deployments have led to burnout and deteriorating morale inside the agency.
Mental stress no limited to evaluation companies
Similar reports have come out from major tech companies, including OpenAI, Anthropic and Google DeepMind, where several specialists have resigned after concluding their projects carry excessive societal danger.The report notes that a Google DeepMind staffer described the effort to establish safety safeguards as an uphill battle that feels like an impossible task, noting that working on AI carries the strain of not knowing how deployments will ultimately be utilised. The worker cited Google’s April agreement granting the US Department of Defense (DOD) access to its systems for classified defense applications as a major source of internal tension.“So many people are upset, there have been various quittings and a lot of mental health leave. Partly this feeling of being out of control, and also that you’re building something that is causing real damage,” the employee was quoted as saying.Insiders argue that these workplace stresses stem from corporate rivalry and management decisions, complicated by the introduction of autonomous systems into sensitive domains like power grids and conflict zones without adequate legal or commercial boundaries, The Financial Times report added.
Bioterror and cyber vulnerabilities spark growing alarm
Some specialists monitoring cybersecurity and biochemistry have voiced concern following sudden advances in algorithmic capabilities, including systems capable of uncovering zero-day software exploits or designing novel biological pathogens. A former Aisi researcher described an atmosphere of pervasive dread within testing teams:“There is this sense that reckless things are being embedded in how tech is released and that is not changing. There is constant anxiety from people who are saying, can anything stop this, or are we watching the world burn?”Responding to the concerns, official bodies and research firms maintained that employee care programs remain active.The UK Government stated that while frontier technological testing is demanding and fast-paced, employee wellbeing and manageable workflows remain central to Aisi’s operations, with support services open to affected personnel.Google DeepMind, which is the AI unit of Google, acknowledged the high-intensity nature of advanced research, asserting that worker wellness resources, formal avenues to escalate safety concerns, and direct communication channels with executive leadership remain top priorities.Recently, Jacob Coxon, a 27-year-old researcher, who previously worked at both OpenAI and Anthropic, resigned publicly this month while accusing top labs of gambling with human survival.Mrinank Sharma, Anthropic safety specialist focusing on mitigating AI-enabled bioterrorism left in February, cautioning in a departure note that the world faces severe peril.Hieu Pham, formerly with OpenAI and Elon Musk’s xAI, Pham stepped down earlier this year to return to Vietnam, writing candidly about the severe toll of advanced model development.
