Several staff members at the UK’s AI Safety Institute (AISI) are currently on sick leave and receiving psychological support due to work-related stress. The institute, tasked with evaluating frontier AI models before public release, is experiencing high levels of burnout among its cybersecurity and biochemistry teams, who are facing AI’s growing ability to discover software vulnerabilities and generate novel biological threats. In May 2026, a restructuring merged the societal resilience team into the human impacts unit, reducing staff from approximately 15 researchers to just three. Andrew Strait, who headed the societal resilience team, resigned in July 2026. Established in November 2023 and rebranded in February 2025, AISI operates on a voluntary evaluation basis and lacks statutory authority to compel companies to submit their models for testing or block a model’s release.
Source: Read the original article
