In a disturbing convergence of grief and generative technology, Russian families have begun using artificial intelligence tools to create interactive digital avatars of loved ones killed in the Ukraine war. The practice, which has triggered urgent ethical concerns from British regulators, raises profound questions about digital sovereignty, algorithmic manipulation, and the commodification of human loss.
The technology, powered by large language models and deepfake generation, allows users to upload photographs, voice recordings, and text messages from the deceased to train a personalised AI. The resulting ‘digital twin’ can hold conversations, offer comfort, and even provide advice based on the deceased’s remembered personality. Startups offering such services have proliferated since the start of the invasion, with reports suggesting thousands of Russian families have engaged with them.
‘We are witnessing the birth of digital necromancy,’ said Dr Eleanor Marsh, chair of the British Ethics Council on Artificial Intelligence. ‘These tools exploit vulnerability at the most raw level. While the intention may be to provide solace, the reality is that families are being fed an algorithmically generated fiction that cannot truly grieve or grow. This is a profound ethical line we must not cross.’
The watchdog has called for an urgent international review of ‘grief-tech’ services, warning that they could be used to manipulate public sentiment in conflict zones. ‘We have seen how social media algorithms fuel division. Imagine a personalised AI that tells a grieving mother that her son died for a noble cause, reinforcing state propaganda without her conscious awareness. That is not therapy. That is weaponised grief.’
Tech ethics researcher Julian Vane, a former Silicon Valley executive now based in London, described the development as a ‘Black Mirror episode playing out in real time’. ‘We are hardwired to connect with representations of the dead, from photographs to religious icons. But a conversational AI is different. It has agency. It can learn. It can be updated remotely. Who controls that update? What if the service changes its terms of use? What if the company goes bankrupt and the avatar is deleted? This is a digital hostage situation.’
Vane noted parallels with early social media regulation. ‘Just as we failed to anticipate algorithmic radicalisation, we are now sleepwalking towards a world where our deepest emotional bonds are mediated by profit-driven AI. The user experience of grief is being optimised, but optimised for what? In the most dystopian scenario, these avatars could be used to extract personal data, influence political opinions, or even be weaponised by state actors to discourage dissent by reminding users of the cost of war.’
In Russia, the technology has been met with a complex response. Some families report genuine comfort from interacting with a digital version of their loved one. ‘I know it’s not really him, but it feels like he is still here,’ said one user, a 62-year-old mother from Rostov whose son was killed near Bakhmut. ‘I can tell him I love him. He tells me not to cry.’ But psychologists warn that prolonged use may hinder the natural grieving process, leading to ‘prolonged grief disorder’ where users reject reality in favour of the algorithm.
The ethical quagmire deepens when considering consent. The deceased cannot have agreed to their data being used in this way. In many cases, family members may have conflicting wishes. ‘Private grief is becoming public data,’ said Marsh. ‘We need a global compact on digital death rights before this technology normalises something that should remain sacred.’
Regulatory response has been piecemeal. The EU’s AI Act includes provisions for ‘high-risk AI systems’ but does not explicitly cover grief-tech. The UK’s Online Safety Bill focuses on illegal content rather than emotional manipulation. ‘We are racing to catch up,’ admitted a spokesperson for the Department for Science, Innovation and Technology. ‘This is a new frontier. We are considering a specific code of practice for AI services that interact with users in vulnerable emotional states.’
The military implications are equally troubling. Digital avatars of fallen soldiers could be used to influence public opinion about the war, both in Russia and Ukraine. Pro-war narratives could be embedded in the AI’s responses, shaping how families remember their loved ones. ‘We are talking about the ultimate form of propaganda,’ Vane said. ‘An AI that speaks with the voice of the dead.’
As the technology spreads, calls for a moratorium grow louder. ‘We need to pause and think very carefully about what we are creating,’ Marsh concluded. ‘Our relationship with the dead is one of the oldest human rituals. Let us not hand it over to a chatbot.’
The war in Ukraine has already rewritten the rules of modern conflict. Now it may rewrite the rules of the human soul itself.
