In a chilling display of technology outpacing morality, Russian families are turning to artificial intelligence to create digital avatars of soldiers killed in the Ukraine war. The practice, which involves feeding photographs, videos, and voice recordings into generative AI models, produces eerily lifelike chatbots that mimic the dead. Britain’s National Cyber Security Centre has issued a stark warning about the ethical and psychological risks, calling it a “dangerous frontier” where grief meets unchecked innovation.
The technology, powered by deep learning and natural language processing, allows families to text or speak with a virtual reconstruction of their loved one. Startups in Russia and beyond now offer these services for a fee, capitalising on a market of trauma. But experts say the digital resurrection raises profound questions about consent, mental health, and the weaponisation of memory.
Julian Vane, Technology & Innovation Lead at the Institute for Digital Futures, comments: “We are witnessing the commodification of grief. These avatars are not the person; they are stochastic parrots trained on private data. The families are being sold a simulation of closure, but what they get is a feedback loop that can deepen trauma. Worse, the data used could be siphoned for surveillance or propaganda, turning private mourning into a state asset.”
The British government has urged caution, pointing to the lack of regulation around “grief tech”. Officials note that while the AI models may offer comfort in the short term, they could hinder the natural grieving process and create dependency. There are also concerns about data sovereignty: who owns the digital persona of a deceased soldier, and could it be used to spread disinformation or recruit?
This incident is not isolated. Similar services have emerged in South Korea and China, where AI chatbots for the bereaved are a growing industry. But the context of war adds a geopolitical edge. Russian families, desperate for connection amid state propaganda that often glosses over casualties, may be vulnerable to manipulation. Meanwhile, Ukraine has condemned the practice as an exploitation of death for political ends.
The moral calculus is tricky. For centuries, humans have sought to commune with the dead through séances or photographs. AI offers a more interactive illusion, one that is ethically ambiguous. As Vane notes: “We must ask whether every technological possibility should be realised. The fact that we can build a digital ghost does not mean we should. User experience here is not about convenience; it is about human dignity.”
Britain’s warning may be a harbinger of future regulation. The government is exploring a “Digital Afterlife Bill” that would require explicit consent for generating AI personas after death, and mandate rigorous psychological safeguards. Until then, the burden falls on individuals to recognise that talking to a machine is not the same as talking to the lost. In the end, the dead do not speak. Only the algorithms do.
