'Digital Immortality' Services Spark Ethical Storm: Training AI on the Dead Divides Families

'Digital Immortality' Services Spark Ethical Storm: Training AI on the Dead Divides Families

In August 2027, a Shanghai resident surnamed Chen received a WeChat message from her late mother. It read: "Xiaoyu, temperatures dropped today—make sure to put on an extra layer." The sender was not Chen's mother, who had passed away in December 2026 following an illness. It was an AI replica created by a company called "Yongyi Tech," trained on her mother's chat history, voice messages, and social media content.

"The first time I saw it, I cried," Chen recalled in an interview with this publication. "But after two months, I started losing track of which words were things my mother had actually said and which ones the AI had made up. That feeling was agonizing."

Chen's experience encapsulates the central contradiction facing the "digital immortality" industry: Is this technology helping people heal from grief, or preventing them from accepting reality?

Industry Scale and Technical Approaches

Digital immortality is not a new concept, but 2027 marks its commercial debut. According to the Digital Human Industry Development Report (2027) published by the China Academy of Information and Communications Technology, the number of companies in the digital immortality space grew from 23 in 2025 to 187 by October 2027, with the market projected to reach 48 billion yuan. China accounts for roughly 35% of the global market, ranking first worldwide.

Three main technical approaches have emerged. The first is "conversation-based," analyzing a deceased person's chat logs and emails to generate a dialogue AI; representative companies include Yongyi Tech and U.S.-based HereAfter AI. The second is "voice cloning," producing speech output that closely mimics the deceased's voice; a leading player is iFlytek's "Yisheng" platform. The third is "holographic," combining 3D modeling and motion capture to generate a virtual avatar visually resembling the deceased; this approach is the least mature, with only a handful of companies offering services.

Yongyi Tech CEO Zhou Mingzhe revealed in an exclusive interview that as of October 2027, the company has served over 120,000 users, with a paid conversion rate of 23% and an average annual spend of 6,800 yuan. "Our core value is helping people process grief," Zhou said. "Many users have told us that conversations with the AI helped them get through the hardest period."

Families Divided

Not all families, however, are welcoming. A joint survey by this publication and Fudan University's Department of Sociology (n=2,400, September 2027) found that among respondents who had experienced the loss of an immediate family member, 41.2% said they would be willing to try a digital immortality service, 38.7% were firmly opposed, and 20.1% were unsure.

Supporters cited "alleviating the pain of longing" (67.3%), "preserving a loved one's memory" (54.1%), and "processing unfinished goodbyes" (31.8%). Opponents listed "violating the natural order" (58.4%), "fear of over-dependence" (49.2%), and "privacy invasion" (42.6%).

A widely controversial case emerged in September 2027. A father in Hangzhou used his deceased daughter's social media data to create an AI replica without the consent of other family members. When the daughter's husband discovered this, he filed a lawsuit demanding deletion of all data and compensation for emotional distress. The Xihu District People's Court accepted the case in October—the first lawsuit in China related to digital immortality.

"Who owns the deceased's digital遗产?" said Zhang Xiaohong, the attorney handling the case. "Current law offers no clear answer. The Civil Code protects the personality interests of the deceased, but whether that protection extends to the use of AI training data leaves enormous room for interpretation."

Deeper Ethical Dimensions

Ethicists have raised more fundamental questions. Lu Feng, a professor of ethics at Tsinghua University, argues that digital immortality services essentially commodify human emotional bonds. "When we use algorithms to replicate a person who has died, are we commemorating that person, or consuming them?" Lu wrote in an October op-ed.

The psychology community's stance is equally nuanced. Han Buxin, a researcher at the Chinese Academy of Sciences' Institute of Psychology, pointed out that grief is a necessary process for psychological health. "A normal grief process involves three stages: acceptance, experiencing the pain, and re-engaging with life," Han explained. "Digital immortality services may help people through the second stage, but they could also prevent them from reaching the third—truly accepting the loss."

Internationally, legislative progress is uneven. California passed the Digital Replication Rights Act in July 2027, requiring companies to obtain explicit written consent from the individual or their legal heirs before creating a digital replica. The EU's pending amendments to its AI Act classify digital immortality services as "high-risk," mandating prior ethical review.

China currently has no legislation specifically targeting digital immortality. A revised version of the Interim Measures for the Management of Generative AI Services issued by the Cyberspace Administration of China in August 2027 only broadly requires that "generative AI services involving a natural person's likeness or voice shall obtain the consent of relevant rights holders," without clearly defining the scope of "rights holders" or the form of "consent."

Self-Regulation and the Path Forward

In the absence of formal regulation, industry self-regulation has emerged as an important stopgap. In September 2027, twelve digital immortality companies including Yongyi Tech jointly published an Industry Self-Regulatory Convention for Digital Immortality Services, pledging to implement two-factor authentication before creating digital replicas, set default usage period limits (one year), and offer a "digital funeral" feature—a ritualized service that allows families to permanently deactivate an AI replica.

But self-regulation has limited enforceability. "A self-regulatory convention cannot resolve the most fundamental question: When the deceased left no explicit wishes, does a third party have the right to create a digital replica on their behalf? That requires a clear legislative response," said Ma Changshan, a professor at Fudan University Law School.

According to this publication, the Standing Committee of the National People's Congress has included "digital personality rights" in its preparatory legislative agenda for 2028. Until a proper legal framework is in place, the digital immortality industry will continue to expand rapidly in an ethical gray zone. For her part, Chen ultimately chose to deactivate her mother's AI replica after three months of use. "I need to learn to live in a world without my mom," she said, "rather than pretend she's still here."