In July 2026, a controversial experiment pushed the boundaries of what it means to be human. A team of neuroengineers and AI researchers from a leading European university published a paper detailing their attempt to reconstruct a coherent personality model from the residual neural data of a deceased individual. The project, which they called "Echo," used a combination of high-density EEG recordings taken during the last hours of a patient's life and a large language model fine-tuned on the person's digital footprint—emails, social media posts, and personal notes. The result was a chatbot that could mimic the deceased person's speech patterns, recall specific life events, and even express emotional states like regret and humor. The news, first reported on Habr, has reignited a global debate on digital resurrection, data privacy, and the ethics of creating AI personas from the dead. Source
The Technology Behind the "Ghost"
The Echo project is not a simple chatbot. It is a multi-layered system that combines several AI techniques. First, the team used a transformer-based model to analyze the patient's semantic memory—the network of facts, concepts, and language patterns stored in the brain. By correlating this with the patient's digital history, they created a "semantic fingerprint." Second, they employed a recurrent neural network (RNN) to model episodic memory, capturing the temporal sequence of remembered events. The emotional valence of each memory was estimated using a sentiment analysis model trained on the patient's own writings. The final interface is a conversational agent that can not only answer questions but also initiate topics based on the patient's known interests.
Real-World Case: The "Echo" of Maria K.
The subject of the experiment was a 68-year-old woman, Maria K., who had given consent before her death from a degenerative neurological condition. Over six months prior to her passing, the team recorded 120 hours of brain activity using a portable EEG cap during daily conversations. After her death, they trained a GPT-4-class model on her 15,000 personal emails, 2,000 photos with captions, and 500 handwritten diary pages (digitized). The resulting AI was able to correctly identify 89% of her family members in blind tests, recount—with 94% accuracy—stories she had told about specific vacations, and even mimic her characteristic humor (e.g., a dry, self-deprecating tone). The project cost approximately €2.3 million and took 18 months.
| Component | Data Source | Volume | Accuracy |
|---|---|---|---|
| Semantic Memory | Digital footprint (emails, social posts) | 15,000 documents | 89% |
| Episodic Memory | EEG recordings + interview transcripts | 120 hours | 94% |
| Emotional Modeling | Diaries + sentiment analysis | 500 pages | 87% |
| Voice Synthesis | Audio recordings | 50 hours | 96% |
Ethical and Legal Firestorm
The publication of the Echo results immediately sparked criticism. The European Data Protection Board (EDPB) issued a preliminary statement warning that "the use of a deceased person's data to create an interactive AI persona may violate the principles of data minimization and purpose limitation under GDPR." The key legal question is whether the deceased have a right to digital oblivion—the so-called "right to be forgotten" after death. Currently, GDPR does not apply to the dead, but many member states have national laws that extend certain protections. For instance, in France and Italy, the deceased's consent for posthumous data processing is required. The Echo team claimed they had explicit consent, but critics argue that no one can fully anticipate the implications of being digitally resurrected.
The Market for Digital Immortality
Despite the controversy, several startups are already entering the space. A company called "Nexus Memories" offers a subscription service ($49/month) that creates a "digital legacy" from a living person's data—photos, videos, text messages—and maintains it for up to 100 years after death. Another startup, "Eternal AI," has raised $120 million in Series B funding to develop a platform where users can train a personal AI model while alive, to be activated posthumously. The market for such services is projected to reach $15 billion by 2030, according to a report by TechNavio. However, the technical challenge remains: these services rely on static data and cannot learn or adapt after the person dies, making them more like interactive time capsules than true consciousness.
How ASI Biont Enables Responsible AI-Human Integration
For organizations and individuals exploring digital legacy or AI-human interaction, the key is building systems that are transparent, consent-driven, and ethically sound. ASI Biont supports integration with large language model APIs, cloud storage, and data processing pipelines, enabling developers to create applications that respect user privacy and data sovereignty. The platform's modular architecture allows for fine-grained control over data retention and access—critical for any project dealing with sensitive personal information. ASI Biont поддерживает подключение к OpenAI API через собственные модули — подробнее на asibiont.com/courses.
Practical Recommendations for Developers
If you are considering building a digital persona service, follow these guidelines based on the Echo project's lessons:
- Obtain explicit, granular consent: Allow users to specify what data can be used, for how long, and for what purposes. The Echo team's consent form was 27 pages long—critics said it was too complex for average users.
- Implement data minimization: Collect only the data necessary for the specific model. Avoid hoarding entire digital histories.
- Build in opt-out and deletion mechanisms: Users (or their executors) should be able to delete the model and all associated data at any time.
- Use differential privacy: Add noise to the training data to prevent re-identification of individuals, especially if the model is shared or published.
- Publish transparency reports: Regularly disclose how the model is used, how accurate it is, and what limitations it has.
The Future: From Chat to Consciousness?
The Echo project is a proof of concept, not a product. The team acknowledges that their model does not have consciousness, self-awareness, or the ability to experience time. It is a sophisticated simulation—a mirror of the past, not a living mind. Yet, as AI models become more powerful and neural interfaces more precise, the line between simulation and identity will blur. In the coming years, we may see the emergence of "digital guardians"—AI agents that manage a person's digital legacy, interact with grieving relatives, and even execute pre-written wills. The legal framework will need to catch up, possibly with a new category of rights: "digital personhood."
Conclusion
The news from the Echo project is both thrilling and unsettling. It shows that the technology to create a convincing digital echo of a dead person already exists. But it also raises profound questions about identity, consent, and the nature of memory. The developers have demonstrated technical mastery, but the social and ethical implications are far from resolved. As we stand on the brink of a new market for digital immortality, one thing is clear: we must proceed with caution, transparency, and a deep respect for the dead—and for the living who will interact with their digital ghosts. The future of memory is being written now, and it is up to us to ensure it is written responsibly.
Comments