What the First AI Guidelines for Older Adults Mean for Caregiving

If you’ve ever cared for an aging parent, you know the paperwork alone can feel like a part-time job. You’re managing a stack of medical bills, tracking five different prescriptions, and trying to decipher visit summaries from three different specialists. It’s completely understandable that families and healthcare providers are looking for help. Generative AI tools are stepping in to summarize medical records, draft messages, and even act as conversational companions. But when we invite a computer program into something as deeply personal as eldercare, we need rules.

Until recently, there were no specific medical guidelines for using artificial intelligence in geriatrics. That changed in July 2026, when the American Geriatrics Society (AGS) published the first-of-its-kind position statement on generative AI in the care of older adults. Published in the Journal of the American Geriatrics Society, this framework tackles the ethical, clinical, and practical realities of using AI with seniors.

This isn’t a technical manual for software developers. It’s a line in the sand for how we treat our elders as technology changes. Whether you’re a family caregiver, a nonprofit director, or a community advocate, these new AI guidelines for older adults matter to you. Let’s look at what the AGS recommends and how it impacts the way we care for the seniors in our lives.

Augmenting, Not Replacing, Human Care

The loudest message in the AGS position statement is a simple one: AI should support human judgment, not replace it. The society emphasizes that generative AI must be integrated responsibly, acting as an assistant rather than the primary decision-maker.

Think about a goals-of-care discussion. When an older adult is deciding whether to pursue an aggressive treatment or transition to comfort care, they aren’t just looking for a statistical probability. They are looking for empathy, eye contact, and a deep understanding of their personal values. An AI tool might be able to quickly pull a patient’s medical history from a massive database, but it can’t read the hesitation in a senior’s voice or understand the nuanced family dynamics in the exam room.

The AGS insists on “clinician-in-the-loop” oversight. This means that anytime an AI model suggests a diagnosis, treatment plan, or medication change, a human doctor or nurse must review it. For community organizations and family caregivers, this translates to a clear boundary. You can absolutely use AI to help organize notes or look up general information, just like we teach in our workshops led by Micah Berkley. But you must never let the tool make the final call on a medical or caregiving decision. Relational care—the actual human connection between a patient, their family, and their doctor—must always stay front and center.

Why Training Data Matters for Seniors

One of the most fascinating and urgent points raised by the AGS involves how AI systems actually learn. Generative AI tools are trained on massive amounts of data pulled from the internet and medical records. But here’s the problem: older adults are frequently underrepresented in those training datasets.

When a computer model doesn’t have enough information about a specific group of people, it struggles to give accurate advice for them. Geriatric care involves highly specific challenges that don’t typically apply to a healthy thirty-year-old. For example, older adults often face polypharmacy, which is the medical term for taking multiple prescription medications at the same time. Managing how these different drugs interact is incredibly complex. If an AI tool is trained mostly on data from younger patients who take one or two medications, it might suggest a treatment plan that is actively harmful to an eighty-year-old taking seven different pills.

The same goes for cognitive assessments and evaluating a patient’s functional status, meaning their ability to perform daily tasks like bathing, dressing, and eating. The AGS highlights that Alzheimer’s disease and related dementias require specialized AI evaluation that accounts for cognitive impairment and decision-making capacity. Because these crucial aspects of eldercare are often high-stakes and nuanced, the guidelines warn against over-relying on automated recommendations. The AI simply might not have the right background knowledge to understand the complexities of an aging body and mind.

This data blindspot is exactly why we advise caution when families start experimenting with consumer AI tools. If you’re Using AI to Understand Medical Bills, the technology is operating in a fairly straightforward, administrative way. But if you ask a chatbot for medical advice regarding a complex mix of age-related conditions, you’re taking a massive risk. The AGS is calling on developers to specifically validate their AI systems in older adult populations using age-relevant outcomes before those tools are deployed in clinics.

Privacy and the Right to Know

When you sit in a doctor’s office, you expect that your conversation is private. You assume your health data is protected. But as generative AI enters the picture, those assumptions need to be explicitly protected by new rules. The AGS framework places a heavy emphasis on transparency and privacy protections across the entire AI lifecycle.

Transparency means that older adults and their care partners have a fundamental right to know when artificial intelligence is being used in their care. If a doctor is using an AI tool to draft a visit summary, the patient should be told. If an automated system is flagging a chart for a potential medication adjustment, that process shouldn’t happen in secret. Trust is the foundation of healthcare. If seniors feel like they are being quietly managed by algorithms, they will lose trust in their providers.

Privacy is equally critical. Generative AI systems often require vast amounts of personal data to function effectively. The AGS warns that without strong governance, there is a real risk of privacy harms. Health systems and tech developers must establish strict safeguards to monitor for unintended consequences, algorithmic bias, and data breaches. They have to continually audit these tools before, during, and after they are deployed.

For those of us working in community nonprofits, this is a loud reminder to audit the tools we use. Whether you’re running a senior center or managing a volunteer caregiving network, you have to protect the data you collect. You might be exploring AI Companions for Older Adults to help combat loneliness, but you have to ask hard questions about where the conversational data is going. Who owns it? Is it being used to train future models? The American Geriatrics Society is telling us that convenience can never come at the cost of a senior’s dignity or privacy.

Bringing These Standards to Our Communities

The release of the American Geriatrics Society position statement is a massive step forward. It takes abstract ideas about technology and translates them into actionable guidance for doctors, health systems, and policymakers. But these guidelines shouldn’t just stay in a medical journal. They need to inform how all of us interact with the older adults in our lives.

As generative AI continues to spread, you are going to see more software pitched at caregivers. You’ll see apps that promise to manage schedules, track symptoms, and analyze behavior. Some of these tools will be genuinely helpful. Others will be built by companies that don’t understand the first thing about person-centered care.

Your job as an advocate is to ask the right questions. Does this tool have a human in the loop? Was it tested on actual older adults, or just on younger populations? Is it transparent about how it uses personal data?

We have a responsibility to bridge the digital divide carefully. As we navigate the AI for Seniors conversation, we have to prioritize equity and shared decision-making. We can embrace technology that makes caregiving a little bit easier and gives doctors more time to look their patients in the eye. But we must stubbornly refuse any tool that tries to automate away the empathy, respect, and deep human connection that older adults deserve.

— Marcus Brown