AI news for business · checked November 17, 2026

On 16 November 2026 OpenAI added computer-use agents, improved clinical search and more than 20 new plugins to ChatGPT for Healthcare. What this means for healthcare organizations.

In healthcare, information is spread across the patient record, hospital policies, scheduling systems and payer portals. On 16 November 2026 OpenAI introduced three additions to ChatGPT for Healthcare for this: agents that operate existing software, improved clinical search, and more than 20 new plugins. The facts come from two official OpenAI posts from the same day; the assessment for organizations in Germany is ours.

LOL.MARKETING / CONNECTING INTELLIGENCE

The essential answer

According to OpenAI, computer-use agents can carry out multi-step work in existing healthcare software by reading screens, clicking, typing and moving between applications. ChatGPT for Healthcare customers get these agents in ChatGPT Work and Codex, along with the improved clinical search, starting now, subject to workspace settings; administrators enable the new plugins. For organizations, OpenAI names role-based access controls, audit logs, configurable approvals, and that content is not used for training. HIPAA-compliant workflows require a Business Associate Agreement with OpenAI. The free ChatGPT for Clinicians is for verified clinicians in the U.S.

01 / FIT

A good fit when

  • Hospitals, care networks, payers and healthcare software vendors that use or evaluate ChatGPT for Healthcare and want to automate recurring administrative workflows across several systems with human approval.

02 / LIMITS

Not the first choice when

  • Organizations without a clear legal basis for using health data with AI, and clinicians outside the U.S. hoping for the free ChatGPT for Clinicians; OpenAI names that offer only for verified clinicians in the U.S.

What OpenAI introduced on 16 November 2026

OpenAI describes the problem this way: healthcare teams have to piece together information from the EHR, hospital policies, scheduling systems and payer portals. ChatGPT for Healthcare is meant to bring existing tools and medical evidence together on one platform. On 16 November 2026 three building blocks were added: advances in computer-use agents, improved trusted clinical search that finds, evaluates and cites medical evidence, and more than 20 new plugins.

According to OpenAI, ChatGPT for Healthcare is a governed workspace for clinical, research and administrative teams. It is distinct from ChatGPT for Clinicians, which is free for eligible, verified clinicians in the U.S. According to OpenAI, both products offer a Business Associate Agreement to support HIPAA-compliant workflows.

Computer-use agents in existing software

Computer-use agents operate software the way people do: they read screens, click, type and move between applications. This is meant to let them take on steps that no complete interface covers, such as moving documents between systems or processing referrals. As an example, OpenAI shows an agent that reviews the chart, physical therapy records and prior imaging, drafts an MRI order, an encounter note and a patient reply, and prepares a prior authorization request in the insurer’s portal, each for human review. The demo uses synthetic patient data.

OpenAI cites its own evaluations: in a review of EHR inbox tasks, 99 percent of 325 reviewed tasks with GPT-6 Astra were rated safe, and physicians rated 93 percent of 646 assessments as very or extremely helpful. On the independent HealthAdminBench from Stanford researchers, GPT-6 Astra earned 94.3 percent of available points according to OpenAI, and GPT-6 Luna 91.8 percent at 98 percent lower cost of 8.5 U.S. cents per task. These are vendor figures, not our own measurements.

Clinical search and new plugins

Trusted clinical search in ChatGPT for Healthcare and ChatGPT for Clinicians is meant to find, evaluate and cite medical evidence better, with citations that can be inspected. In a blind evaluation with over 1,300 reviews, physicians rated the updated experience more highly across all eight dimensions evaluated, according to OpenAI, including factual correctness, safety, actionability and source authoritativeness.

More than 20 new plugins from companies across clinical care, care coordination and business operations are available in ChatGPT for Healthcare according to OpenAI, including through ChartGPT in the EHR workflow. Examples include UpToDate Expert AI, Redox, Metriport, Luma Health, Stedi and Trizetto. Typical tasks according to OpenAI: reviewing patient records and hospital protocols, preparing prescriptions for clinician review, coordinating patient outreach, checking insurance eligibility and analyzing costs. In our assessment, many of these plugins target U.S. systems, for example for insurance checks.

Controls and availability

For secure deployments, OpenAI names role-based access controls and audit logs in ChatGPT for Healthcare. Organizations can set which applications agents can access and which steps require human review, and build in additional approval steps. According to OpenAI, content shared with ChatGPT for Healthcare is not used to train models. For HIPAA-compliant workflows with protected health information, organizations must use eligible services covered by their BAA with OpenAI and the required security and retention settings.

On availability: ChatGPT for Healthcare customers get computer-use agents in ChatGPT Work and Codex and the improved clinical search starting now, subject to workspace settings; administrators enable new plugins. Developers can find computer use in the OpenAI API, ChatGPT Work and Codex. ChatGPT for Healthcare can be purchased directly through sales or via an AWS Marketplace private offer.

Our assessment for organizations in Germany

The announcement is strongly tailored to the U.S. market: HIPAA and the Business Associate Agreement are U.S. legal frameworks, ChatGPT for Clinicians is free only in the U.S., and many of the named plugins connect U.S. systems. The two posts do not give a country list for ChatGPT for Healthcare. Anyone planning in Germany should ask OpenAI directly about availability and contract terms.

Under the GDPR, health data is a special category of personal data. Before agents work with patient data, you need a clear legal basis, a suitable data processing agreement, a data protection impact assessment and clear rules on which steps always require human approval. It makes sense to start with an administrative workflow that involves no treatment decision, such as moving documents, with a complete audit log. This is our assessment, not a statement by OpenAI and not legal advice.

Keep it verifiable

Primary sources

The next sensible step

Is your AI use with health data properly governed?

Eight steps from a general interest in AI to a clearer decision for your business.

Start AI System Check
FreeProvider-neutralNo credentials