AI adoption widens; accountability tightens

OpenAI and Britain’s pharmacy regulator moved in the same week to put artificial intelligence deeper into care — with humans still on the hook for the result. (openai.com) (pharmacyregulation.org) On April 22, OpenAI said ChatGPT for Clinicians is now free for verified U.S. physicians, nurse practitioners, physician assistants, and pharmacists. The product is built for individual clinicians, not hospital-wide deployment. (openai.com) (help.openai.com) OpenAI said the tool is meant for evidence review, documentation, and medical research, and includes clinical search with citations, pre-built workflows, starter prompts, deep research, and continuing medical education support on eligible questions. Sign-up requires a ChatGPT account, a valid National Provider Identifier, and license verification through a third-party provider. (help.openai.com) Two days earlier, on April 20, the General Pharmaceutical Council published a position statement saying pharmacists, pharmacy technicians, and pharmacy owners must meet the same standards when they use artificial intelligence as when they do not. The regulator said professionals remain personally accountable for decisions and actions and must use their own judgment when AI tools are involved. (pharmacyregulation.org) The council also said pharmacy staff should be transparent about AI use, protect patient confidentiality, and explain to patients how AI is being used in their care, including risks and benefits. It published separate advice for education, training, and revalidation alongside the main statement. (pharmacyregulation.org) The backdrop is faster adoption inside medicine. OpenAI cited a 2026 American Medical Association survey showing 72% of physicians reported using AI in clinical practice, up from 48% a year earlier. (openai.com) (ama-assn.org) OpenAI has been building the institutional side of that market too. In January, it launched OpenAI for Healthcare and said ChatGPT for Healthcare was rolling out to health systems including AdventHealth, Baylor Scott & White Health, Boston Children’s Hospital, Cedars-Sinai, HCA Healthcare, Memorial Sloan Kettering, Stanford Medicine Children’s Health, and the University of California, San Francisco. (openai.com) That split matters in practice: individual clinicians can now get a free tool tied to their own license verification, while hospitals that need centralized controls and a Business Associate Agreement are pushed to the enterprise product. OpenAI’s help documentation says ChatGPT for Clinicians is for individual use, and directs organizations seeking admin controls or multi-user coverage to ChatGPT for Healthcare. (help.openai.com) (openai.com) The regulator’s statement in Britain points the same way on governance. It does not ban AI in pharmacy; it says the professional duty stays with the pharmacist or pharmacy technician, even when software drafts, summarizes, or suggests an answer. (pharmacyregulation.org) The result is a clearer operating model for vendors selling AI into care settings in 2026: faster access, more cited outputs, and more workflow support — alongside verification, disclosure, privacy safeguards, and a named professional who still has to sign off. (help.openai.com) (pharmacyregulation.org)