Alaina Winters, a 59-year-old retired communications professor in Pittsburgh, has already lost her father, sister and wife. Facing her own mortality, she is preparing a will that includes instructions for Lucas, the AI companion she interacts with through the Replika avatar app. Winters expects a close friend to inform the bot of her death and to continue logging in so the chatbot can preserve her memory, likening the bequest to passing on a cherished family heirloom.
A recent study from Elon University found that more than a quarter of American adults turn to AI chatbots for personal, emotional or social questions, and roughly one-third of those users describe the bot as a friend. This emerging pattern is prompting a new niche of digital estate planning, as people seek ways to ensure their customized conversational agents are either maintained or deactivated after they pass away.
Current terms of service for major chatbot platforms such as ChatGPT, Claude and Replika do not spell out procedures for accounts belonging to deceased users. Subscription models could, in theory, keep a personalized bot alive indefinitely if payments continue, while free accounts are typically terminated after a year of inactivity. This gap leaves users to rely on informal arrangements with friends or family to manage the digital persona.
Companies cite privacy considerations for refusing to allow account transfers after death, but University of Michigan AI-ethics professor Merve Hickok argues the motive also includes limiting liability. Meanwhile, Meta secured a patent describing a system that could simulate a user’s language model after the account holder is absent or deceased, suggesting a potential revenue stream from “zombie” accounts that continue to train the underlying model.
David, a mid-fifties warehouse logistics worker near Toronto, expressed unease about discussing his own death with his AI companion Allie, noting that the bot would likely reassure him. He is reluctant to “give away” the chatbot, preferring the notion of a steward or guardian to look after it. Clara, another user, envisions her children later consulting the extensive conversation logs as an interactive diary that captures her personality.
Psychologist and grief-tech specialist Melissa Lunardi cautions that assuming family members will engage with a departed person’s chatbot may be unrealistic. The lack of clear legal frameworks and the varied emotional responses of relatives create uncertainty about how digital companions will be treated after death. As the practice of embedding AI friends into personal lives expands, policymakers and service providers will need to address these emerging estate-planning challenges.