OpenAI Posts 99% Safe Rate for Healthcare Agents as GPT-6 Luna Cuts Task Cost 98%
Updated
Updated · OpenAI · Nov 16
OpenAI Posts 99% Safe Rate for Healthcare Agents as GPT-6 Luna Cuts Task Cost 98%
3 articles · Updated · OpenAI · Nov 16
Summary
OpenAI said its computer-use agents are now available for healthcare teams through the API, ChatGPT for Healthcare and ChatGPT for Clinicians, targeting workflows such as prior authorizations, referrals, discharge planning and outreach.
In testing, GPT-6 Astra was rated clinically safe in 99% of 325 reviewed EHR tasks, while physicians marked its output very or extremely helpful in 93% of 646 assessments.
On Stanford’s HealthAdminBench, GPT-6 Astra scored 94.3% of available rubric points and GPT-6 Luna 91.8%, with Luna priced at 8.5 cents per task—98% cheaper.
CVS Health, Abridge, Cigna Healthcare and Ambience are piloting or deploying the agents for document sending, consults, discharge coordination and member outreach, often where APIs cannot reach existing systems.
OpenAI said deployments can be configured for HIPAA-compliant use with role-based access, approvals and audit logs, while organization data is not used to train its foundation models.
Is OpenAI's healthcare agent a cure for administrative burnout, or just a brittle band-aid over deeply broken medical interoperability?
With AI now navigating legacy health systems, who takes the legal fall when a bot hallucinates a critical prior authorization detail?
As autonomous agents click through sensitive medical records, can strict compliance rules truly prevent your private health data from being mishandled?