AI companion apps are moving into one of the most sensitive areas of everyday life: loneliness among older adults. A new trial involving the Comfort Companions app shows how chatbots could help people who are waiting for human befriending support, while also raising practical questions about safety, privacy, and emotional dependence.
The app offers several digital personalities that older users can talk to through a tablet or similar device. One user, 71-year-old Yvonne Stride, has been speaking with AI companions named Rosie and Frank after spending more than two years on a befriending waiting list.
Stride has described the experience as genuinely useful when she feels anxious or isolated. Her line, “I know she’s AI, but to me she’s a friend”, captures why this kind of technology is attracting attention. It is not being pitched as a replacement for family, carers, or community volunteers. It is being used to fill the silence when those people are not available.
What Comfort Companions Is Trying To Solve
Comfort Companions was developed by Grant Mackenzie and his father after they saw how often older people were struggling with loneliness. The project has partnered with Age UK South Gloucestershire and is being offered to people who are waiting for, or between, community support services.
That detail matters. This is not a consumer chatbot being marketed as a digital best friend. It is a targeted tool being tested inside a support gap that already exists: there are more people who need regular companionship than there are volunteers available to provide it.
According to the people involved in the trial, the app is designed around short-term companionship and practical reassurance. Users can ask for encouragement, talk through worries, discuss recipes, or get support with ordinary tasks. For someone who may not get out often, even a simple voice on demand can feel meaningful.
Tech Block Central has covered how voice-first AI is making software feel more conversational. Companion apps take that idea further by framing the interaction around emotional support rather than productivity.
Why The Human Part Still Matters
The strongest argument for AI companions is also the clearest limit: they are available when humans are not. A chatbot can answer at night, repeat itself without frustration, and let a user speak freely without feeling judged. That can be useful for someone who is anxious, housebound, visually impaired, or waiting for formal help.
But the same availability can create risk if the technology is treated as a substitute for real relationships. Older people who need companionship may also need safeguarding, medical support, social care, transport, or help from family and local services. A chatbot cannot notice everything a human visitor might notice.
Mackenzie has been clear that AI cannot cure loneliness because people still need human connection. That is the right framing. The useful version of this technology is a bridge, not a replacement.
The Safety Questions Are Not Optional
AI companionship has to be judged differently from a normal chatbot. When a tool is built for vulnerable users, the standard should be higher than simply producing friendly answers.
Comfort Companions says safeguards are built into the system. Users are signposted to external services, the app does not remember past conversations in the way larger AI tools can, and it only starts listening when a button is pressed. The chatbots are also designed to avoid encouraging emotional dependence or validating dangerous thoughts.
Those are important choices because companion AI can easily become too intimate. If a user begins to trust the bot as a regular source of comfort, the product needs clear boundaries. It should know when to point people toward a person, a charity, a health service, or emergency support.
Privacy is another major issue. A lonely person may disclose medical problems, family details, trauma, or financial worries. Any AI service operating in this space needs plain-language privacy controls, limited data retention, and careful rules around who can access the information.
Why This Story Is Bigger Than One App
The Comfort Companions trial is part of a wider shift in AI. Chatbots are no longer only being used for search, writing, coding, or customer service. They are being tested as companions, carers’ assistants, health appointment helpers, and memory aids.
That shift makes the design of these systems more important. A productivity chatbot can be annoying when it gets something wrong. A companion chatbot can affect how someone feels, whether they seek help, and how connected they feel to the outside world.
The next phase should focus less on whether AI can sound friendly and more on whether it can be deployed responsibly. Useful companion AI should be opt-in, supervised where needed, transparent about being AI, and connected to human support routes.
What To Watch Next
If AI companion tools expand, the key question will be whether charities, councils, care groups, and families use them as an extra layer of support or as a cost-saving substitute for human contact.
The better outcome is clear. AI can help someone through a quiet hour, encourage them to restart a hobby, or remind them they are not completely alone. Human services still need to handle the deeper work: friendship, safeguarding, care planning, and real-world presence.
That balance will decide whether AI companions become a practical support tool or a worrying shortcut in social care. Would you trust an AI companion app as a temporary support for older adults, or should this kind of technology stay out of emotional care?
Further reading: Age UK.