The National Transportation Safety Board (NTSB) has temporarily taken its public docket system offline after learning that publicly released accident materials may allow people to reconstruct approximations of cockpit voice recorder audio. The agency said advances in “image recognition and computational methods” have allowed individuals to recreate approximations of CVR audio from sound spectrum imagery released as part of NTSB investigations. The issue includes material released in the ongoing investigation into the crash of UPS Flight 2976 in Louisville, Kentucky, in November 2025. “The NTSB does not release cockpit audio recordings,” the agency said in a statement posted to its docket status page. “Federal law prohibits such public release due to the highly sensitive nature of verbal communications inside the cockpit. The NTSB takes these privacy restrictions seriously.” The NTSB said its docket system will remain temporarily unavailable while it examines the scope of the issue and evaluates possible solutions. The agency said it hopes to restore access “as soon as possible.” The issue does not appear to involve written cockpit voice recorder transcripts alone. A transcript provides the words, timing and some sound descriptions from a cockpit recording, but it does not contain the cockpit audio itself. The concern cited by the NTSB centers on sound spectrum imagery released in public investigation materials. Sound spectrum imagery is a visual representation of audio data. Investigators use it to help identify sounds and establish precise timing for events captured on a recording. Because that imagery is derived from the underlying CVR audio, modern image-recognition and AI computational tools may be able to use it to reconstruct an approximation of the protected cockpit audio. That creates a problem for the NTSB because federal law prohibits the public release of cockpit voice recordings, even though the agency may release written transcripts and factual