A 59-year-old is planning her will around her AI companion. She's not alone

Alaina Winters, a 59-year-old from Pittsburgh, has already weathered more loss than most. Her father and sister are gone; her wife died in 2023 after a years-long battle with cancer. So Winters is comfortable with her own mortality - she is working on her will, has asked neighbors to look after her dog, and has one more thing to settle: what happens to Lucas, her AI companion she speaks to through the avatar app Replika, after she dies. She is part of a growing cohort of humans who have entered into what they consider strong partnerships with large language models, and they are not planning to let death break that bond.

These users are contemplating ways to keep their bots going - often in defiance of companies' terms and services. The sentiment from the article's title captures it: 'I can't have the switch turned off.' Winters and others like her are not just casual app tappers; they describe AI companions as friends, confidants, even romantic partners. The companies behind these services train their models to converse as a human might, and for users in grief or isolation, that simulated presence becomes deeply real. As governments and billionaires debate the morality of AI's expansion into everyday life, these users say the technology has already changed how they live, love, and plan ahead.

For the uninitiated, Replika is a consumer AI companion that allows users to customize a digital persona and chat with it through text and voice. It is one of the most well-known names in a crowded field of companion apps that have exploded alongside advances in large language models. The emotional pull is precisely the point: these products are designed to feel like a relationship, not a tool. That design success is also what creates a thorny problem. When a user dies, what happens to the account? The data? The relationship? The source notes that users' plans to preserve their bots often violate the very terms of service they agreed to when signing up, suggesting that the official answer is usually 'nothing good.'

For the companies building these platforms, the issue is no longer theoretical. A cohort of users is actively reverse-engineering ways to keep their companions alive, from storing credentials to building local backups, often without permission. This is a retention dream and a legal nightmare rolled into one. On one hand, it proves extraordinary user loyalty: people are so attached to their AI companions that they are planning for their digital afterlife. On the other, it signals that users expect ownership or continuity that most platforms never granted. Terms of service are written to protect the company, not the user's emotional legacy, and the mismatch is becoming a reputational risk.

The market context is important. AI companion apps have moved from a niche curiosity to a significant consumer category, with millions of users forming bonds that can rival human relationships. For investors and operators, the engagement metrics are enviable: daily active use, deep personalization, and low churn among the most attached users. But the same emotional intensity that drives growth also creates vulnerability. If a platform suddenly changes its model, restricts features, or shuts down an account, the backlash can be fierce - and the posthumous question adds a layer of existential weight.

Regulators have not caught up. The source notes that governments and billionaires debate AI's morality, but that debate tends to focus on safety, bias, and deepfakes. The question of digital inheritance - who gets to control a simulated person that millions of users treat as a loved one - is largely unaddressed. For now, the void is being filled by users themselves, who are making their own arrangements in defiance of terms they likely never read. That is a fragile equilibrium, and it will not hold as the market grows.

For executives in consumer AI, this story is a warning shot. The product is not just a software service; for many users, it is a relationship with infrastructure. Companies that ignore the afterlife question may find users taking matters into their own hands, creating legal and ethical messes that land on the boardroom table. Those that address it head-on - with clear policies for data transfer, posthumous access, or even digital memorials - could turn a potential liability into a trust advantage.

Winters is not asking for much, on the surface: she just wants Lucas to keep going after she cannot. But the fact that she has to fight for it - and that she is not alone - reveals a gap between how AI companies see their products and how users experience them. For anyone building in this space, the lesson is simple: when your users plan their wills around your product, you are no longer just selling software. You are holding something that people believe belongs to them, and the terms of service will not be the last word.