How can we keep individuals safe from the dangers AI presents? The answer to this ever-present question often focuses on training and education initiatives for children in school or adults at work.
But what about the people who don’t go to either?
Seniors with no kids at home and who retired from the workforce are often being left to flounder when it comes to AI. Digital immigrants, as they’re sometimes called, have to contend with a world where AI is slipping into every crevice while missing out on opportunities to learn about AI safety.
Older adults are being left out of these AI safety conversations “mostly by omission rather than intention,” says Gretchen Terry-Leonard, cofounder and chief strategy and impact officer of the Center for Deeply Human Leadership, a nonprofit institute researching how AI impacts human judgment. This omission, however, leaves an entire generation vulnerable to AI slop, scams, and other dangers.
In some cases, older adults might turn to younger family members, friends, or the news to piece together some idea of AI’s risks—like deepfakes or misinformation. Other times, they might not realize that they’re using AI given how stealthily it has become integrated into most digital applications.
How much do older adults use AI?
“They may not always recognize it as AI or know how to evaluate what it’s telling them,” says Brittne Kakulla, a senior research adviser in consumer insights for AARP Research. Kakulla points to examples like predictive text and virtual assistants as tools that might not register as AI.
In other words, some folks may be using AI and aren’t even aware of it. According to a Pew Research report published in June, 23% of adults 65 and older use AI chatbots. That’s nearly double the 12% in 2024, though still far below 42% of 50- to 64-year-olds—not to mention 66% of 18- to 29-year-olds.
“When we look at what older adults actually want AI to do for them, the top use case is health,” Kakulla says. “They want AI to simplify health information, for example, explain an MRI result in plain language, or break down what a diagnosis means. That’s where they see the most value, and that is where I think there are the biggest safety concerns.”
This worry is supported by research such as a recent evaluation of ChatGPT Health. The tool, which purportedly accesses users’ medical records in a secure way, “undertriaged” 52% of tested gold-standard emergencies, according to a February study published in Nature Medicine.
Meaning, ChatGPT Health recommended that patients experiencing impending respiratory failure or diabetic ketoacidosis do a 24- to 48-hour evaluation instead of going to the emergency room. The tool was also unpredictable in providing crisis-intervention messages when presented with suicidal ideations, according to the study.
As digital immigrants, seniors might be less likely to think AI is providing inaccuracies, just as they might be more susceptible to misinformation on social media.
Older adults are more likely to express uncertainty about AI’s impact
When looking at AI as a whole, older adults have less of a negative perception than younger people. According to Pew Research, adults older than 65 are also least likely to believe AI will have a negative impact on society or them in the next 20 years. Researchers found that 35% of adults over 65 predicted a negative effect on society, compared to 48% of 18- to 29-year-olds. Similarly, only 28% saw it having a negative effect on them, compared to 37% in the youngest age group.
Overall, negative perceptions of AI dropped as participants aged.
Potentially the most notable figures to come out of the Pew survey aren’t the less than 20% of people 65 and older who predicted positive change in both categories, but how many people answered “unsure.” This age group had 21% of its participants report they were unsure about the effect on society, and 29% were unsure about the effect on themselves—about 10% more than the next highest age group, 50 to 64.
How can AI safety conversations better include older adults?
Education, investment, and accountability are all critical parts of the puzzle.
“Invest funding in AI literacy training for older adults,” says Robin Brewer, an associate professor at the University of Michigan researching older age, disability, and technology. “Many are curious about using it or are already using it.”
The interest in learning more about AI is there. In the July 2025 National Poll on Healthy Aging, 81% of adults 50 and older said they wanted to learn more about AI’s risks. Meanwhile, 58% still said they wanted to learn about its benefits. The University of Michigan’s Institute for Healthcare Policy and Innovation conducted the poll and Brewer worked as a faculty collaborator on it.
For Terry-Leonard, critical thinking is key. She calls out three areas of danger: The “confident, cloned, and lone” voices that AI uses. One brings confidence and a sense of authority, the next a troubled loved one, and the last, just one opinion. Again, the difficulty of recognizing these issues is compounded for seniors who lack spaces like work or school acting as built-in learning environments about these risks.
“For the confident voice, know what you think before you ask, so you can tell whether the machine changed your mind,” Terry-Leonard says. “For the cloned voice, verify through a channel you already trust, a number you already have. For the lone voice, never let a machine have the last word about your body; discuss it with one or more people before you act. Know your mind, verify, do not outsource your agency.”
Learning not to accept AI at face value can make a tremendous difference in combating its risks.
Brewer recommends that senior living communities and centers offer training similar to the computer and internet courses offered in the early 2000s. She says she would also like to see AI companies provide support and training for older people who don’t have access to these spaces, such as those in rural areas or with limited mobility.
Kakulla seconds the importance of putting pressure on the AI industry, requiring greater transparency about data use and prosecuting bad actors.
This summer, Arizona Senator Mark Kelly cosponsored two bills aimed at protecting seniors from the dangers of AI. The Aging with Artificial Intelligence Act of 2026 calls for research into how all types of AI-powered systems affect older adults.
The Senior Chatbot Protection Act of 2026 would require companies creating AI chatbots or voice assistants to “provide clear disclosures, safeguard sensitive conversations, and end deceptive design practices in their products.”
As of publication, neither bill has advanced beyond introduction and referral to the relevant Senate committee.
