NTSB temporarily removes docket access after AI-generated pilot voices from crash emerge
L'essentiel
- The NTSB has temporarily disabled its docket system after discovering AI-generated recreations of deceased pilots' voices from a UPS crash last year were circulating online.
- The voices were reconstructed from a spectrogram file of the cockpit voice recorder, using AI tools.
Résumé généré par IA
Pourquoi c'est important
The National Transportation Safety Board (NTSB) is responsible for investigating transportation accidents. Their docket system historically provides public access to investigation data. Federal law prohibits including cockpit audio recordings in the docket.
In the latest sign of these AI-heavy times, the National Transportation Safety Board temporarily removed access to its docket system after discovering that voices of pilots who were killed in a UPS plane crash last year had been recreated using AI and were circulating on the internet.
NTSB is prohibited by federal law from including cockpit audio recordings in its docket system, which otherwise contains troves of data on investigations and has historically been open to the public. But the accident docket for this flight included a spectrogram file of the voice recorder. A spectrogram uses a mathematical process to turn sound signals, including low and high frequencies, into an image.
Scott Manley, a popular YouTuber channel who combines physics, astronomy, and video games, noted on X that it could be possible to reconstruct audio from the megabytes of data encoded in that image.
And that’s what happened. People took the spectrogram, along with the publicly available transcript, to create approximations of the cockpit voice recorder audio from UPS flight 2976 in Louisville, Kentucky, according to the NTSB. They used AI tools like Codex, according to posts on social media.
Questions ouvertes
- How many individuals accessed the spectrogram file?
- What specific AI tools were used by individuals to reconstruct the audio?
- What is the NTSB's plan to prevent future unauthorized access or misuse of data?
- What is the extent of the circulation of the AI-generated voices?






