Signal of the Day · Security & AI Risk Surfaced late · published August 12
CHAI convenes ~100-member Health AI Cybersecurity Work Group, playbooks due end of 2026
The Coalition for Health AI announced a nearly 100-member Health AI Cybersecurity Work Group meeting bi-weekly to produce the sector's first peer-developed frontier-model cyber guidance: a defensive playbook, an offensive playbook, and a frontier AI cyber-risk assessment tool, with the playbooks targeted for release at the end of 2026. CHAI's announcement names a specific frontier-model capability jump as having compressed attack timelines and industrialized protected health information exfiltration and ransomware — a standards body treating the threat model as having changed on a date rather than gradually. The fourteen-seat leadership council spans academic medical centers, community health, payers and security vendors; no skilled-nursing, senior-living or post-acute organization holds a seat, so SNF and post-acute operators should plan to consume these outputs rather than shape them, and watch whether the risk-assessment tool is scoped for organizations without a dedicated security operations center.