AI is being used to resurrect the voices of dead pilots
The spread of AI voice-cloning technology has produced an unexpected and unsettling use case: internet users are recreating the voices of pilots who died in plane crashes, drawing on the cockpit voice recordings (cockpit voice recorder transcripts and audio segments) contained in crash-investigation documents. According to TechCrunch's report, this content is spreading rapidly on social-media platforms — particularly in short-video formats — as 'dramatised retellings' of crashes.
The technical basis of the practice rests on generative voice-synthesis tools that have advanced rapidly over the past three years. Models that can clone a person's voice from a few seconds of audio sample are now widespread as free and easy-to-use tools. TechCrunch's reporter notes that these tools produce content reconstructing 'the final moments of a crash' based on the real pilot voices in crash recordings. While this content is often based on real transcripts, it is sometimes extended with fictional dialogue.
At the centre of the ethical debate is the question of recreating the voices of the deceased without their consent. For the families of pilots who died in aviation accidents, this content can be especially distressing. TechCrunch reports that some families have sent takedown requests to social-media platforms but that the platforms' responses have been inconsistent. The legal status of the voice rights of the deceased is not clearly defined in most countries — an ambiguity that complicates content-removal processes.
The public availability of crash-investigation documents supplies the raw material for the practice. In the US, the National Transportation Safety Board (NTSB) and the aviation-safety agencies of other countries publish documents and, in some cases, audio recordings for the transparency of crash investigations. This transparency is critical to improving aviation safety; but in the AI era, new and unforeseen uses of these documents are emerging. An NTSB spokesperson told TechCrunch that 'the public availability of crash documents is a cornerstone of the aviation-safety culture, but we are concerned about disrespectful uses of these documents.'
Aviation-safety experts warn that this content may undermine the serious nature of crash analysis. Independent aviation-safety consultant John Cox told TechCrunch that 'crash investigations are rigorous, scientific studies conducted to prevent future crashes; turning these recordings into sensational content is both disrespectful to the victims and damaging to the aviation-safety culture.' Cox emphasised that the original purpose of cockpit voice recordings is 'not blame but learning.'
Platform responsibility is another dimension of the debate. Social-media platforms' content policies have not yet fully settled their rules on AI-generated audio content. Some platforms have introduced a requirement to label 'AI-generated content', but the enforcement of these rules is inconsistent. TechCrunch reports that content creators can easily circumvent these rules and that algorithms' tendency to promote emotionally intense content accelerates the spread of such material.
From a broader perspective, the AI 'resurrection' of the deceased is not limited to the aviation field. Practices such as using musicians' voices for new songs after their deaths, 'bringing back to life' historical figures as AI chatbots, and families creating AI avatars of lost loved ones have become widespread in recent years. Each of these practices is part of a broader legal and ethical debate about how digital voice and image rights should be managed after death.
The regulatory framework lags behind these developments. Some US states — particularly Tennessee's ELVIS Act, which protects musicians — have introduced legal protection for the voice and image rights of the deceased. But there is not yet a comprehensive framework at the federal level or internationally. Legal experts note that existing copyright and personality-rights laws cannot fully respond to the new problems created by the AI era.
TechCrunch's report is an example of how the democratisation of technology — powerful AI tools accessible to everyone — brings unforeseen ethical problems with it. Voice-cloning technology can be applied across a wide spectrum, from legitimate uses (accessibility, dubbing, restoration of lost audio) to harmful uses (fraud, impersonation, non-consensual content production). Although the technology itself is neutral, drawing the ethical limits of its use will remain an important societal debate in the years ahead.
This article is not investment or legal advice; for personal or institutional decisions on AI voice technologies and digital rights, consultation with relevant legal experts is recommended. TechCrunch said it will track how platforms' policies on AI-generated audio content evolve in the coming months.