AI is being used to resurrect the voices of dead pilots

2026-05-22 · Source: AI News & Artificial Intelligence | TechCrunch · Field: Legal & Regulatory — Regulatory Affairs & Government Relations, Compliance & Risk Management · Depth: Novice, quick

Summary

The National Transportation Safety Board (NTSB) temporarily restricted public access to its docket system after discovering that artificial intelligence tools were used to recreate the voices of pilots killed in a UPS plane crash last year. This incident stemmed from a spectrogram file of the cockpit voice recorder, which was included in the accident docket for UPS Flight 2976 in Louisville, Kentucky. Although federal law prohibits the NTSB from releasing cockpit audio, a spectrogram, which converts sound signals into an image, was publicly available. Individuals, including popular YouTuber Scott Manley, noted the possibility of reconstructing audio from this data. Using AI tools like Codex and publicly available transcripts, approximations of the cockpit voice recorder audio were created and circulated online. The NTSB has since restored public access but kept 42 investigations, including Flight 2976's, closed for review.

Key takeaway

For policy makers and legal professionals drafting data release guidelines, this incident highlights the urgent need to reassess what constitutes "redacted" or "non-audio" data. Your current policies may be insufficient against advanced AI reconstruction capabilities. You should consider how seemingly innocuous data formats, like spectrograms, can be reverse-engineered to reveal sensitive information, necessitating stricter controls on all publicly accessible investigation materials to prevent unauthorized voice recreation.

Key insights

AI tools can reconstruct sensitive audio from publicly available spectrograms, posing privacy and ethical challenges.

Principles

• Spectrograms contain reconstructible audio data.
• Federal law restricts cockpit audio release.
• Public data can be repurposed with AI.

Method

Individuals used a publicly available spectrogram file and transcript, combined with AI tools like Codex, to approximate cockpit voice recorder audio.

In practice

• Review data formats for unintended audio leakage.
• Assess AI's capacity to reconstruct sensitive data.
• Implement stricter data redaction policies.

Topics

• AI Voice Reconstruction
• Spectrogram Analysis
• NTSB Data Policy
• Data Privacy
• Ethical AI Use
• Cockpit Voice Recorders

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Ethicist, Policy Maker, Legal Professional

Related on AIssential

• US scrambles to stop Internet users re-creating dead pilots’ voices — Spectrograms combined with AI tools enable unauthorized audio reconstruction, challenging privacy laws.
• AI Was Used to Recreate the Voices of Dead Pilots. The NTSB Responded by Locking Down Its Database. — Spectrograms contain reconstructible audio data, enabling AI voice synthesis and prompting NTSB database lockdown.
• ElevenLabs hit with fresh lawsuit over use of voices by Pulitzer and Emmy-winning journalists — AI voice synthesis companies face significant legal risks for unauthorized use of copyrighted voice data.
• The DeepSpeak-Agentic Dataset — A new 37-hour video dataset enables forensic identification and interaction studies of embodied AI agents.
• From cloning romance authors to YouTube piracy, AI is transforming audiobooks — AI is rapidly transforming audiobook creation and distribution, introducing both accessibility benefits and significant challenges like piracy and ethical concerns.
• AI Voice Cloning Scams Are Now Draining As Much As $635,000 From Their Victims After Just A 5-Second Audio Sample From A Loved One — AI voice cloning, requiring minimal audio, exposes critical vulnerabilities in voice-based identity verification systems across industries.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI News & Artificial Intelligence | TechCrunch.