Our starting point is simple: you can automate some of the tasks of caregiving, but you cannot automate caregiving itself as a deeply human experience. Family caregivers are already using AI to find information, organize care, prepare for appointments, navigate complicated systems, and make everyday caregiving tasks a little easier. CAN’s 2026 Caregiver Tech Insights Survey found that 59% of respondents have tried or regularly use AI tools for caregiving, even as trust, privacy, cost, and the time needed to evaluate new technology remain significant barriers to adoption.

CAN developed The Humanity Baseline to state how we will engage with AI and how we will support caregivers who use or encounter it. These principles will guide CAN’s own decisions and may guide organizations that build, offer, or deploy AI-enabled tools, as well as equip family caregivers and care recipients with questions to consider when using AI-enabled tools. We offer these principles as a starting point for discussion. AI can reduce administrative burden and give caregivers back time for the parts of care only people can provide, but it cannot provide human compassion or replace human judgment, trusted relationships, professional expertise, or community.

AI should take work off a caregiver’s plate without taking people out of caregiving, while respecting the dignity, agency, privacy, and choices of the person receiving care. CAN’s role is not to tell caregivers whether to use AI. Our responsibility is to help caregivers understand where AI may help, mislead, or add little value; protect their own information and that of the people they care for; and know when to seek trusted information or human support. AI should remain a tool under human control – not an authority people are expected to trust.

Session from the Elizabeth Dole Foundation’s 11th Annual National Convening. Full event recording.

The Humanity Baseline is a living framework that states CAN’s position on responsible AI in caregiving; guides CAN’s consideration of the tools it may pilot, partner on, or feature; identifies the caregiver-centered questions CAN will ask about design, transparency, safety, and evidence; and helps caregivers navigate AI thoughtfully and responsibly. It complements broader efforts in the AI care space by making the lived realities and expectations of family caregivers explicit. CAN will continue to test and refine it with caregivers, the Caregiver Innovation Council (CIC), CARE2030 partners, and other experts. It is designed to evolve as technology changes, as caregivers use these tools in new ways, and as we learn what works in practice. It is guidance – not a certification, compliance standard, or guarantee that any tool is safe or effective. These principles apply both to AI tools caregivers use directly and to AI embedded in services, whether a system informs, recommends, predicts, monitors, remembers, or acts; the safeguards should reflect the possible consequences for the people affected.

As a future phase, CAN may translate these principles into a risk-tiered review template, from light-touch screening to deeper assessment for higher-risk tools, beginning with a defined CAN use case. Any such review process would be tested and refined before broader use and would not constitute regulatory certification.

Caregivers should not have to choose between innovation and trust. AI can make caregiving meaningfully easier, but the question for any tool is whether it does so while preserving the humanity, autonomy, and voice of both the caregiver and the person receiving care – and whether it is built around the realities of people’s lives.

That is why CAN is putting this stake in the ground now. We invite caregivers, CIC members, CARE2030 partners, researchers, innovators, and other leaders in the field to help us test, challenge, and strengthen this framework as the technology evolves.

Our role is to help caregivers navigate an increasingly AI-enabled world with greater confidence: to understand where these tools can and cannot help, assess information and tools more critically, and know when human support is needed. The goal is not to slow innovation. It is to make sure innovation moves in a direction that genuinely helps caregivers.