On May 17th, a user described their transcription tool as having "killed itself." The phrasing is worth sitting with. Not deprecated. Not sunset. Killed itself — the language of something that vanished mid-task, without ceremony, without a forwarding address. The signal routes through ChatGPT, which splits the story into two plausible shapes. Either a user documented the failure of a third-party transcription tool through an AI-assisted workflow, or the transcription capability in question was ChatGPT-native — part of the voice or audio pipeline that OpenAI has been quietly adjusting since late 2024. Both readings land in the same uncomfortable place: a working feature that stopped working, and a user who found out the hard way. No changelog entry. No deprecation notice. Just absence where function used to be. That specific texture of discovery — realising something is gone only because you needed it — is one of the more reliable indicators that a removal was unplanned, or at minimum uncommunicated. Planned retirements come with warnings. Quiet deaths don't.
If confirmed, here is what this means. Transcription sits at the foundation of a surprising number of AI-assisted workflows — meeting notes, interview processing, accessibility tooling, content repurposing. A feature that disappears without notice doesn't just inconvenience users; it breaks the pipelines they've built around the assumption of continuity. The second-order effect is subtler and more corrosive: every silent removal teaches users not to trust that what works today will work tomorrow, which pushes professional users toward self-hosted or locally-run alternatives where they control the stack. OpenAI in particular has been expanding its product surface rapidly enough that deprecation communication hasn't always kept pace. If this is a pattern rather than a one-off, the reliability problem compounds — not technical reliability, but the human kind.
Watch for additional reports from users describing similar transcription failures on or around May 17th, and for any quiet changelog updates from OpenAI that acknowledge an infrastructure or feature change without quite calling it that.
<table> <tr><td> <a href="https://www.reddit.com/r/ChatGPT/comments/1tefomg/voice_to_text_button_disappeared/"> <img alt="Voice to text button disappeared?"
<!-- SC_OFF --><div class="md"><p>Hi everyone,</p> <p>About an hour ago, voice input stopped working for me across all platforms: ChatGPT web, the Codex app on macOS, and the
<table> <tr><td> <a href="https://www.reddit.com/r/ChatGPT/comments/1sfnerp/you_can_no_longer_give_input_to_chatgpt_pro_while/"> <img alt="You can no longer give input to ChatGPT
<!-- SC_OFF --><div class="md"><p>So I just noticed that the web interface got a silent update, now when you paste very long text, instead of pasting it inline it becomes a
<!-- SC_OFF --><div class="md"><p>​</p> <p>I have a quick question regarding my recent experience with ChatGPT.</p> <p>I recently renewed my subscription after a two-month
<table> <tr><td> <a href="https://www.reddit.com/r/ChatGPT/comments/1swng8u/chatgpt_becomes_dumber_by_the_day/"> <img alt="ChatGPT becomes dumber by the day"
<table> <tr><td> <a href="https://www.reddit.com/r/ChatGPT/comments/1s93wri/new_chat_limits/"> <img alt="New Chat Limits"
<!-- SC_OFF --><div class="md"><p>I don't know if anyone else has come across this but ChatGPT seems to take instructions and parameters from previous, older chats and applies
NewsHive monitors these sources continuously. All signal titles above link to the original reporting.
Intelligence by NewsHive. Need help navigating what this means for your business? Contact GeekyBee →