> ## Content Index > Fetch the complete content index at: https://www.thedigitalspeaker.com/llms.txt > Use this file to discover other available public pages before exploring further. # Synthetic Minds | AI Has Gone From Advising to Prescribing Medication - URL: https://www.thedigitalspeaker.com/synthetic-minds-ai-advising-prescribing-medication/ - Published: 2026-07-03T06:22:24.000Z - Updated: 2026-08-04T05:42:31.000Z - Description: An FDA-cleared AI agent has begun altering patients' medication inside major US hospitals, the government funds one that manages heart care, and algorithms already shape what insurers cover. The AI in medicine has moved from advising the human decision to making it. No one has settled who is liable. - Author: Dr Mark van Rijmenam, CSP - Tags: Synthetic Minds Newsletter, #newsletter *The Synthetic Minds newsletter offers short daily insights to get you thinking. If you enjoy it, please forward. All signals are powered by* [*Futurwise*](https://futurwise.com/?ref=thedigitalspeaker.com)*. If you need more insights, subscribe to Futurwise and *get 25% off* for the first three months!* ***I have just launched the*** [***Intelligence Age Scorecard!***](https://www.thedigitalspeaker.com/intelligence-age-scorecard/) ***It will help you understand how ready your organization is for the Intelligence Age.*** **Today’s topic:* Health & Longevity* --- ### [When Your Prescriber Is Software, Who Is Liable?](http://thedigitalspeaker.com/synthetic-minds-ai-advising-prescribing-medication/?ref=thedigitalspeaker.com) An [AI](https://www.thedigitalspeaker.com/ai-strategy-speaker/) is adjusting the insulin of patients inside the Cleveland Clinic. A doctor set the limits, but the software carries the change forward. It is FDA-cleared, and it is live. The [AI](https://www.thedigitalspeaker.com/ai-speaker/) in medicine has stopped advising the human decision and started making part of it. A patient-facing AI agent talks to people and [adjusts their insulin ](https://www.futurwise.com/article/a3592354-87d5-48f2-b813-3648f2c7aeeb?ref=thedigitalspeaker.com)within limits a physician sets, deployed at Cleveland Clinic, Allegheny Health Network, and UCSF Health. The cleared device is bounded; a narrow insulin tool wrapped in conversation. The precedent is not bounded at all. The government has [funded the sequel](https://www.fiercehealthcare.com/ai-and-machine-learning/trump-administration-creating-clinical-ai-agents-3-year-fda-approval?ref=thedigitalspeaker.com): the first agent authorized to adjust heart medication around the clock, on a multi-year path to approval. On the payment side, the rule that [algorithms may help decide what care gets covered](https://www.aamc.org/advocacy-policy/washington-highlights/cms-addresses-use-ai-medicare-advantage-plans?ref=thedigitalspeaker.com) was already settled a few years ago. Prescription at one end, coverage at the other. The software has a hand on both ends. That's the efficiency story. Here is the signal. The AI in medicine has spent a decade as the quiet assistant. It read the scan, drafted the note, flagged the clot. A human always decided. That arrangement has ended, and the announcements dressed the ending as a product launch. An agent that acts is a different machine from an agent that advises. When it adjusts a dose, it operates where a mistake causes physical harm. That raises a question the advisory era never had to answer. Who answers for what the software does to a patient? The federal briefing on the heart-care agent names the gap the press releases skip: responsibility among the developer, the supervising clinician, and the hospital is unsettled. No one has signed for it. Two costs hide behind the convenience. An always-available, confident agent invites doctors and patients to trust it exactly where trust is most dangerous. And a handful of vendors become load-bearing infrastructure for care, so a model update becomes a safety event rather than a support ticket. Aviation learned this order the hard way. The autopilot flew, the captain kept the title, and it took a generation of accidents to settle accountability when the two disagreed. Medicine has put the autopilot in the exam room and skipped that argument. The machine that frames the choice while the human signs for it has reached the bedside. The question is no longer whether the machine is accurate. It is who is accountable the first time it is confidently wrong, and whether anyone signed for that before the agent went to work. --- ## The Intelligence Age Scorecard [![](https://storage.ghost.io/c/af/cc/afcca743-e1e6-4752-bf81-782fb033f39c/content/images/2026/05/intelligence-age-scorecard-copy.jpg)](https://www.thedigitalspeaker.com/intelligence-age-scorecard/) An FDA-cleared AI agent has begun adjusting patients' medication inside major health systems, and the accountability for its actions is unsettled. That is a WAVE question: are you still watching this shift, or should your clinical-governance and risk teams already be adapting to a world where the prescriber is partly software? Benchmark your readiness for the next two quarters with the [Intelligence Age Scorecard](https://www.thedigitalspeaker.com/intelligence-age-scorecard/). Or read the public Intelligence Age Scorecard of [IBM](https://www.thedigitalspeaker.com/ibm-ai-readiness-strong-signals-slower-operating-model/), [Visa](https://www.thedigitalspeaker.com/visa-ai-readiness-pioneering-agent-rules-itself/), [Qantas](https://www.thedigitalspeaker.com/qantas-ai-readiness-announced-outcomes-undisclosed-infrastructure/), [Woolworths](https://www.thedigitalspeaker.com/woolworths-ai-readiness-ambition-thin-spine/), [Telstra](https://www.thedigitalspeaker.com/telstras-ai-readiness-trained-workforce-without-authority-to-move/) or [Commonwealth Bank](https://www.thedigitalspeaker.com/what-cba-public-record-reveals-ai-readiness/) first. --- If this newsletter was forwarded to you, [you can sign up here](https://www.thedigitalspeaker.com/newsletter-archive/). Thank you. Mark ## Frequently asked questions ### What is the AI doing at Cleveland Clinic? A patient-facing AI agent talks to patients and adjusts their insulin within limits a physician sets. It is FDA-cleared and already live, and it has also been deployed at Allegheny Health Network and UCSF Health. This marks a shift from AI merely advising doctors to AI actively carrying out part of the treatment decision. [Link to this question](#faq-what-is-the-ai-doing-at-cleveland-clinic) ### Who is liable if the AI prescribing agent makes a mistake? Responsibility is unsettled among the developer, the supervising clinician, and the hospital. A federal briefing on a related heart-care agent explicitly names this gap, noting that no one has clearly signed for accountability when the software acts and causes harm, unlike the earlier era when a human always made the final call. [Link to this question](#faq-who-is-liable-if-the-ai-prescribing-agent-makes-a-mistake) ### What is the next AI agent being developed after the insulin tool? The government has funded an agent authorized to adjust heart medication around the clock, which is on a multi-year path toward approval. This follows the insulin-adjusting device and signals that AI is moving beyond a narrow, bounded tool into broader autonomous medication management. [Link to this question](#faq-what-is-the-next-ai-agent-being-developed-after-the-insulin) ### Why does it matter that AI can now adjust medication instead of just advising? When an AI acts rather than advises, it operates where mistakes cause physical harm, unlike a decade of AI reading scans or drafting notes with a human deciding. This creates risks: overtrust in a confident, always-available agent, and dependence on a few vendors whose model updates become safety events rather than routine support issues. [Link to this question](#faq-why-does-it-matter-that-ai-can-now-adjust-medication)