record
ChatGPTUse to open an audio recording widget when the user wants to capture microphone input directly inside the ChatGPT conversation. This tool is appropriate when the user asks to record audio, make a voice memo, capture a spoken note, or open a simple recorder UI that they can operate themselves. The tool does not require any input arguments because the recording interaction happens inside the widget after it is displayed. Once opened, the widget lets the user start recording, pause and resume, finish the take, and download the resulting file locally from their browser. The tool is intentionally limited to presenting that recorder interface. It does not transcribe the captured speech, upload audio anywhere, email files, attach the audio to another system, or perform downstream analysis. Do not call this tool for requests that can be answered in plain text, for users who explicitly do not want a widget, or for workflows that require automatic audio processing rather than an interactive recorder. Also avoid using it when the user asks for a different media action such as playing audio, editing existing audio, or generating synthetic speech. Use it specifically when the best next step is to display a microphone recording widget and let the user control recording themselves. The output and metadata only include a timestamp sentinel confirming that the widget invocation completed; the actual audio data remains in the browser and is not returned through the tool response.
