☰
✕
UA RU EN

OpenAI Rolls Out AI Agents in US Healthcare Facilities with Doctors Approving 99% of Tasks as Safe

OpenAI Rolls Out AI Agents in US Healthcare Facilities with Doctors Approving 99% of Tasks as Safe Photo: НВ — Техно

AI Agents Transforming Healthcare Operations Across the United States

OpenAI is introducing AI agents into American medical institutions to streamline both clinical and administrative workflows, significantly reducing time spent on repetitive duties. Healthcare professionals review and finalize the AI-generated outputs, ensuring the technology maintains a high safety standard. Remarkably, physicians have deemed 99% of the tasks completed by these systems as safe.

These AI agents integrate with healthcare software to analyze patient medical histories, prepare documentation, and handle insurance-related inquiries. During a demonstration, one AI agent efficiently processed an MRI request for a patient suffering from lower back pain. OpenAI's partnerships include several major healthcare organizations such as:

  • CVS Health
  • Cigna Healthcare
  • Abridge
  • Ambience

At CVS Health, AI agents assist in sending documents, managing referrals, and preparing consultations. One agent was able to review months’ worth of accumulated records over a single weekend, while another handles software operations for tens of thousands of consultations daily. Cigna Healthcare employs this technology to consolidate data from more than 10 different systems. Abridge develops AI agents to coordinate inpatient care and organize home healthcare services, whereas Ambience uses AI to generate insurance authorization requests.

Ensuring Safety and Maximizing Efficiency with AI

The system is equipped with a fail-safe that halts processes when ambiguous information arises, prompting the need for physician intervention. To validate the technology, OpenAI tested the GPT-6 Astra model on 325 electronic medical record-related tasks, with 99% evaluated as safe. Across 646 individual assessments, doctors rated 93% of the AI’s responses as highly or extremely useful. GPT-6 Astra scored 94.3% on the HealthAdminBench test, outperforming GPT-6 Luna’s 91.8%.

Cost efficiency is another highlight: Astra completes each task at approximately 8.5 cents, making it 98% more affordable compared to Luna. The systems also log screenshots and activity records to monitor AI performance. Healthcare providers can access OpenAI’s API, ChatGPT for Healthcare, and ChatGPT for Clinicians, with customizable agent permissions, human approval workflows, and audit trails to comply with HIPAA regulations.

The integration of AI agents in healthcare settings promises to revolutionize medical service delivery by boosting operational efficiency and reducing expenses. Given the overwhelmingly positive safety and utility feedback from medical professionals, these AI-driven solutions are poised to become new industry standards. OpenAI’s collaboration with leading healthcare organizations highlights a growing commitment to leveraging artificial intelligence to optimize healthcare processes.

While the integration of AI agents in healthcare shows promising results, recent incidents have raised concerns about their oversight. Following two instances of reduced control over these systems, OpenAI has updated its safety protocols to ensure that the technology continues to prioritize patient safety and reliability in medical settings.